Modelcatalogus

We present a preview version of DeepSeek-V4 series, including two strong Mixture-of-Experts (MoE) language models — DeepSeek-V4-Pro with 1.6T parameters (49B activated) and DeepSeek-V4-Flash with 284B parameters (13B activated) — both supporting a context length of one million tokens.

Mistral Small 3.2 24B is een open-source taalmodel van Mistral-small-3.2-24b-instruct-2506 met een contextvenster van 128K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 Embedding 8B is een open-source taalmodel van Qwen met een contextvenster van 33K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3.5 397B A17B is een open-source taalmodel van Qwen3.5-397b-a17b met een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 235B A22B Instruct is een open-source taalmodel van Qwen3-235b-a22b-instruct-2507 met een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 Coder 30B A3B is een open-source taalmodel van Qwen3-coder-30b-a3b-instruct met een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

[!Note] This repository contains model weights and configuration files for the post-trained model in the Hugging Face Transformers format. These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, KTransformers, etc.

Llama 3.3 70B is een open-source taalmodel van Meta met 70B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

We're introducing GLM-5.2, our latest flagship model for long-horizon tasks. It marks a substantial leap in long-horizon task capability over its predecessor GLM-5.1 and, for the first time, delivers that capability on a solid 1M-token context. GLM-5.2's new capabilities include: - Solid 1M Context: A solid 1M-token context that stably sustains long-horizon work - Advanced Coding with Flexible Effort: Stronger coding capabilities with multiple thinking effort levels to balance performance and latency - Improved Architecture: We propose IndexShare, which reuses the same indexer across every fou

We present a preview version of DeepSeek-V4 series, including two strong Mixture-of-Experts (MoE) language models — DeepSeek-V4-Pro with 1.6T parameters (49B activated) and DeepSeek-V4-Flash with 284B parameters (13B activated) — both supporting a context length of one million tokens.

Loes Large World: Qwen3.5-27B dense base met volledig-Europese SFT-adapter (LoRA), geserveerd via vLLM --enable-lora + qwen3 reasoning-parser, CUDA-graphs. Topmodel-spoor voor chat.loes.ai.

Mistral Medium 3.5 is een open-source taalmodel van Mistral-medium-3.5-128b met een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Kimi K2.6 is an open-source, native multimodal agentic model that advances practical capabilities in long-horizon coding, coding-driven design, proactive autonomous execution, and swarm-based task orchestration.

BGE Multilingual Gemma2 is een open-source taalmodel van BAAI met een contextvenster van 8K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen2.5 32B is een open-source taalmodel van Qwen met 32B parameters en een contextvenster van 33K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Llama 3.3 70B Instruct is een open-source taalmodel van Llama-3.3-70b-instruct met een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Mistral Nemo 12B is een open-source taalmodel van Mistral met 12B parameters en een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

GPT-OSS 120B is een open-source taalmodel van Gpt-oss-120b met een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Whisper Large v3 is een open-source taalmodel van OpenAI, gehost op Europese GPU's via een OpenAI-compatibele API.

[!Note] This repository contains model weights and configuration files for the post-trained model in the Hugging Face Transformers format. These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, KTransformers, etc.

Gemma 3 27B is een open-source taalmodel van Gemma-3-27b-it met een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Kimi K2.7 Code is a coding-focused agentic model built upon Kimi K2.6. With substantial improvements on real-world long-horizon coding tasks, it strengthens end-to-end task completion across complex software engineering workflows while improving token efficiency, reducing thinking-token usage by approximately 30% compared with Kimi K2.6.

Gemma 4 26B A4B is een open-source taalmodel van Gemma-4-26b-a4b-it met een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Pixtral 12B is een open-source taalmodel van Pixtral-12b-2409 met een contextvenster van 128K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance. Our approach is built upon three key technical breakthroughs:

The DeepSeek R1 model has undergone a minor version upgrade, with the current version being DeepSeek-R1-0528. In the latest update, DeepSeek R1 has significantly improved its depth of reasoning and inference capabilities by leveraging increased computational resources and introducing algorithmic optimization mechanisms during post-training. The model has demonstrated outstanding performance across various benchmark evaluations, including mathematics, programming, and general logic. Its overall performance is now approaching that of leading models, such as O3 and Gemini 2.5 Pro.

MiniMax M3 is een open-source taalmodel van MiniMaxAI met een contextvenster van 205K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

GLM-5.1 is our next-generation flagship model for agentic engineering, with significantly stronger coding capabilities than its predecessor. It achieves state-of-the-art performance on SWE-Bench Pro and leads GLM-5 by a wide margin on NL2Repo (repo generation) and Terminal-Bench 2.0 (real-world terminal tasks).

Qwen3.6 35B A3B is een open-source taalmodel van Qwen met 35B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Sovereign EU model fine-tuned by HostYourAI on loes-xl-pre.

Qwen2.5 VL 72B is een open-source taalmodel van Qwen met 72B parameters en een contextvenster van 128K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

MiniMax M2.5 is een open-source taalmodel van MiniMaxAI met een contextvenster van 197K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen 3 0.6B is een open-source taalmodel van Qwen met 0.6B parameters en een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Sovereign EU model fine-tuned by HostYourAI on loes-xl-pre.

We are launching GLM-5, targeting complex systems engineering and long-horizon agentic tasks. Scaling is still one of the most important ways to improve the intelligence efficiency of Artificial General Intelligence (AGI). Compared to GLM-4.5, GLM-5 scales from 355B parameters (32B active) to 744B parameters (40B active), and increases pre-training data from 23T to 28.5T tokens. GLM-5 also integrates DeepSeek Sparse Attention (DSA), largely reducing deployment cost while preserving long-context capacity.

Kimi K2.5 is an open-source, native multimodal agentic model built through continual pretraining on approximately 15 trillion mixed visual and text tokens atop Kimi-K2-Base. It seamlessly integrates vision and language understanding with advanced agentic capabilities, instant and thinking modes, as well as conversational and agentic paradigms.

Qwen3 VL 8B is een open-source taalmodel van Qwen met 8B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen 3 4B is een open-source taalmodel van Qwen met 4B parameters en een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen 3 Coder 30B-A3B (MoE) is een open-source taalmodel van Qwen met 30B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Llama 3.2 1B is een open-source taalmodel van Meta met 1B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3.6 27B is een open-source taalmodel van Qwen met 27B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

This is the model card for EuroLLM-22B-Instruct. You can also check the pre-trained version: EuroLLM-22B-2515.

Sovereign EU model fine-tuned by HostYourAI on loes-large-v1.

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

Llama 3.2 3B is een open-source taalmodel van Meta met 3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

medgemma 27b text it is een open-source taalmodel van Google met 27B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Recon2Reason Reasoning 4B is a 4B-scale vision-language model specialized for spatial reasoning in indoor and embodied scenes. It is fine-tuned from Qwen3-VL-4B and improves reasoning about metric distance, relative position, object configuration, and spatial relations from visual inputs.

We are excited to introduce DeepSeek-V4-Flash-Vision-Exp, our first experimental multimodal model in the DeepSeek-V4 family. It builds on the DeepSeek-V4-Flash architecture by incorporating visual modules and undergoing continued training to unlock visual understanding capabilities.

GLM-5.3 uses the same base model as GLM-5.2 — every gain comes from post-training. Compared with GLM-5.2, it is much better at complex coding and long-horizon tasks:

DeepSeek-V4-Pro-0813 is the official release of DeepSeek-V4-Pro, superseding the preview version, with greatly enhanced agentic capabilities and performance improvements that are especially pronounced in production environments. It is built on the DeepSeek-V4-Pro (Preview) model structure, with a DSpark speculative decoding module attached.

Granite is a family of open-source large language models developed by IBM, designed for enterprise and research use. Granite models are built to be versatile, safe, and efficient — covering a range of sizes and capabilities from compact edge-deployable models to large-scale reasoning systems. All Granite models are released under the Apache 2.0 license, enabling unrestricted commercial and academic use.

Granite is a family of open-source large language models developed by IBM, designed for enterprise and research use. Granite models are built to be versatile, safe, and efficient — covering a range of sizes and capabilities from compact edge-deployable models to large-scale reasoning systems. All Granite models are released under the Apache 2.0 license, enabling unrestricted commercial and academic use.

Granite is a family of open-source large language models developed by IBM, designed for enterprise and research use. Granite models are built to be versatile, safe, and efficient — covering a range of sizes and capabilities from compact edge-deployable models to large-scale reasoning systems. All Granite models are released under the Apache 2.0 license, enabling unrestricted commercial and academic use.

Looking for the post-trained model? See NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 for the full-precision release, or NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 for optimized inference.

[!Note] This repository contains model weights and configuration files for the post-trained model in the Hugging Face Transformers format. These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, TokenSpeed, etc.

The pre-training data has a cutoff date of September 2025. The post-training data has a cutoff date of May 2026.

AREX is a family of deep research agents developed by the Beijing Academy of Artificial Intelligence (BAAI). It is designed for long-horizon tasks in which an agent must search across sources, assemble candidate answers, verify multiple constraints, and revise its research plan when the available evidence is incomplete.

AREX is a family of deep research agents developed by the Beijing Academy of Artificial Intelligence (BAAI). It is designed for long-horizon tasks in which an agent must search across sources, assemble candidate answers, verify multiple constraints, and revise its research plan when the available evidence is incomplete.

Fara1.5-27B is a multimodal computer use agent (CUA) for web browsers, from Microsoft Research AI Frontiers. It observes the browser through screenshots and acts on the user's behalf by emitting structured tool calls — click, type, scroll, visit URL, web search, and so on — to complete tasks end-to-end.

Fara1.5-4B is a multimodal computer use agent (CUA) for web browsers, from Microsoft Research AI Frontiers. It observes the browser through screenshots and acts on the user's behalf by emitting structured tool calls — click, type, scroll, visit URL, web search, and so on — to complete tasks end-to-end.

GOVERNING TERMS: Use of this model is governed by the OpenMDW License Agreement, version 1.1. ADDITIONAL INFORMATION: Apache License, Version 2.0.

salamandra 7b fc 2607 is een open-source taalmodel van BSC-LT met 7.8B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

ALIA 40b fc 2607 is een open-source taalmodel van BSC-LT met 40B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Cosmos3-Super-Text2Image-4Step is a 4-step distilled version of the base Cosmos3-Super-Text2Image model. Given a text prompt, it generates a high-fidelity image.

The NVIDIA Qwen-Image-Flash model generates images from text prompts using a four-step, DMD2-distilled version of Qwen/Qwen-Image. The distillation used DMD2 from NVIDIA FastGen, NVIDIA Model Optimizer, and NVIDIA AutoModel while retaining the base model architecture. The packaged scheduler is configured for the four-step, shift-3 trajectory.

[!NOTE] WARNING: Although this model has undergone safety and value alignment, it may still occasionally generate unintended or undesired outputs. Sampling Parameters: For optimal performance, we recommend using temperatures close to zero (0 - 0.2). Additionally, we advise against using any type of repetition penalty, as from our experience, it negatively impacts instructed model's responses.

[!NOTE] WARNING: Although this model has undergone safety and value alignment, it may still occasionally generate unintended or undesired outputs. Sampling Parameters: For optimal performance, we recommend using temperatures close to zero (0 - 0.2). Additionally, we advise against using any type of repetition penalty, as from our experience, it negatively impacts instructed model's responses.

GELab Zero 4B preview Sico Evolution is een multimodaal taalmodel van Microsoft met 4.4B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

[!NOTE] WARNING: This model has been trained on instructions but has not undergone safety or value alignment. Work In Progress: New versions will be released over the coming months.

The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni. The 1.7B version achieves state-of-the-art performance among open-source ASR models and is competitive with the strongest proprietary commercial APIs.

The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni. The 1.7B version achieves state-of-the-art performance among open-source ASR models and is competitive with the strongest proprietary commercial APIs.

The model employs a hybrid MoE architecture with interleaved Mamba, MoE, and Attention layers. Like Nemotron-3-Super, it supports Multi-Token Prediction (MTP) for faster text generation. Compared to its parent, Puzzle-75B-A9B reduces the model from 120.7B total / 12.8B active parameters to 75.3B total / 9.3B active parameters.

Poro 2 Long Instruct is an instruction-following chatbot model with extended context support, created through supervised fine-tuning (SFT) of the Poro 2 Long Base model followed by merging the SFT checkpoint back with the base model to preserve long-context performance. This model is designed for conversational AI applications and instruction following in both Finnish and English, with support for context lengths up to 128K tokens. It was trained on a carefully curated mix of English and Finnish instruction data.

[!Note] This model card is for the new versions of the Gemma 4 family optimized with Quantization-Aware Training (QAT), which allows preserving similar quality to bfloat16 while dramatically reducing the memory requirements to load the model. Four versions of the QAT checkpoints are available: Unquantized QAT checkpoints (Q40): Half-precision weights extracted from the QAT pipeline, ideal for custom downstream compilation and research. Available for Gemma 4 E2B, E4B, 12B, 26B A4B, and 31B, and their drafter models. GGUF (Q40): Ready-to-deploy formats for broad ecosystem compatibility. Availabl

For more details on how to deploy and use the model - see the Quick Start Guide below!

Cosmos3 is a collection of Omnimodal world models capable of generating dynamic, high-quality video, image, audio, and action commands from combinations of text, image, video, and action trajectory inputs. It serves as a foundational building block for a broad range of Physical AI applications and research spanning world understanding, world generation, simulation, and embodied policy learning.

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

[!NOTE] WARNING: This model has been trained on instructions but has not undergone safety or value alignment. Work In Progress: New versions will be released over the coming months.

AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation

AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation

AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation

AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation

MagenticBrain is a 14B-parameter orchestration model from Microsoft Research AI Frontiers. It plans multi-step tasks, calls declared tools, and coordinates sub-agents. It does not execute actions itself — every real-world side effect happens inside a host harness.

Fara1.5-9B is a multimodal computer use agent (CUA) for web browsers, from Microsoft Research AI Frontiers. It observes the browser through screenshots and acts on the user's behalf by emitting structured tool calls — click, type, scroll, visit URL, web search, and so on — to complete tasks end-to-end.

[!NOTE] WARNING: This model has been trained on instructions but has not undergone safety or value alignment. Work In Progress New versions will be available during the coming weeks/months. Sampling Parameters: For optimal performance, we recommend using temperatures close to zero (0 - 0.2). Additionally, we advise against using any type of repetition penalty, as from our experience, it negatively impacts instructed model's responses.

Poro 2 8B Math Reasoning RL Preview is a specialized model focused on mathematical reasoning and problem-solving. This preview model was created through reinforcement learning (RL) on top of the Math Reasoning SFT checkpoint. This model excels at mathematical reasoning tasks but is not optimized for general conversational use or other domains.

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

NVIDIA Cosmos Reason 2 is an open, customizable, 32B-parameter reasoning vision language model (VLM) for physical AI and robotics that enables robots and vision AI agents to reason like humans, using prior knowledge, physics understanding and common sense to understand and act in the real world. This model understands space, time, and fundamental physics, and can serve as a planning model to reason what steps an embodied agent might take next.

[!Note] This model card is for the new versions of the Gemma 4 family optimized with Quantization-Aware Training (QAT), which allows preserving similar quality to bfloat16 while dramatically reducing the memory requirements to load the model. Four versions of the QAT checkpoints are available: Unquantized QAT checkpoints (Q40): Half-precision weights extracted from the QAT pipeline, ideal for custom downstream compilation and research. Available for Gemma 4 E2B, E4B, 12B, 26B A4B, and 31B, and their drafter models. GGUF (Q40): Ready-to-deploy formats for broad ecosystem compatibility. Availabl

[!Note] This model card is for the new versions of the Gemma 4 family optimized with Quantization-Aware Training (QAT), which allows preserving similar quality to bfloat16 while dramatically reducing the memory requirements to load the model. Four versions of the QAT checkpoints are available: Unquantized QAT checkpoints (Q40): Half-precision weights extracted from the QAT pipeline, ideal for custom downstream compilation and research. Available for Gemma 4 E2B, E4B, 12B, 26B A4B, and 31B, and their drafter models. GGUF (Q40): Ready-to-deploy formats for broad ecosystem compatibility. Availabl

Model Summary: Granite-Embedding-311M-Multilingual-R2 is a 311M parameter dense embedding model from the Granite Embeddings collection for high-quality multilingual text embeddings. It produces 768-dimensional vectors with a context length of up to 32,768 tokens. The model supports 200+ languages (based on the multilingual pretraining corpus of the underlying encoder), with enhanced support for 52 languages and programming code that receive explicit retrieval-pair and cross-lingual training. All training data uses permissive, enterprise-friendly licenses, plus IBM-collected and IBM-generated d

Model Summary: Granite-Embedding-97M-Multilingual-R2 is a 97M parameter dense embedding model from the Granite Embeddings collection for high-quality multilingual text embeddings at minimal compute cost. It produces 384-dimensional vectors with a context length of up to 32,768 tokens. The model supports 200+ languages (based on the multilingual pretraining corpus of the underlying encoder), with enhanced support for 52 languages and programming code that receive explicit retrieval-pair and cross-lingual training. All training data uses permissive, enterprise-friendly licenses, plus IBM-collect

Granite Guardian 4.1 8B introduces improved Bring Your Own Criteria (BYOC) support, enabling users to define arbitrary judging criteria beyond the pre-baked safety and hallucination detectors. The model can now faithfully evaluate complex, multi-part requirements such as formatting rules, length constraints, and domain-specific instructions.

Model Summary: Granite Vision 4.1 4B is a vision-language model (VLM) that delivers frontier-level performance on structured document extraction tasks — chart extraction, table extraction, and semantic key-value pair extraction — in a compact 4B parameter footprint, providing a lightweight alternative to much larger frontier models for these tasks:

Granite-Speech-4.1-2B-Plus has similar capabilities to the Granite-Speech-4.1-2B model. The plus model adds two new community-requested rich transcription features that can be activated with a simple prompt change: speaker-attributed ASR (speaker labels and word transcripts) and word-level timing information. Unlike the base mode, the plus model doesn't provide punctuation and capitalization.

Model Summary: Granite Speech 4.1 2B is a compact and efficient speech-language model, specifically designed for multilingual automatic speech recognition (ASR) and bidirectional automatic speech translation (AST) for English, French, German, Spanish, Portuguese and Japanese.

Poro 2 Long Base is an 8B parameter decoder-only transformer created by extending Poro 2 8B Base from an 8K to 128K token context window using LongRoPE. The model supports both English and Finnish with an extended context window of 128K tokens. Poro 2 Long Base is released as a fully open source model under the Llama 3.1 Community License.

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

Using the 🤗's Diffusers library to run URSA in a simple and efficient manner.

Model Summary: Granite-4.1-30B is a 30B parameter long-context instruct model finetuned from Granite-4.1-30B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. Granite 4.1 models have gone through an improved post-training pipeline, including supervised finetuning and reinforcement learning alignment, resulting in enhanced tool calling, instruction following, and chat capabilities.

Model Summary: Granite-4.1-8B is a 8B parameter long-context instruct model finetuned from Granite-4.1-8B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. Granite 4.1 models have gone through an improved post-training pipeline, including supervised finetuning and reinforcement learning alignment, resulting in enhanced tool calling, instruction following, and chat capabilities.

Model Summary: Granite‑4.1‑3B‑Base is a decoder‑only language model with long‑context capabilities, designed to support a broad range of general text‑to‑text generation tasks, as well as fill‑in‑the‑Middle (FIM) code completion. This model shares the same underlying architecture and weights as Granite 4.0 3B Micro, which is trained from scratch on approximately 15 trillion tokens following a four-stage training strategy: 10 trillion tokens in the first stage, 2 trillion in the second, another 2 trillion in the third, and 0.5 trillion in the final stage. An additional training phase is applied

Model Summary: Granite-4.1-3B is a 3B parameter long-context instruct model finetuned from Granite-4.1-3B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. Granite 4.1 models have gone through an improved post-training pipeline, including supervised finetuning and reinforcement learning alignment, resulting in enhanced tool calling, instruction following, and chat capabilities.

EGM-Qwen3-VL-4B-SFT is the supervised fine-tuning (SFT) checkpoint from the first stage of the EGM (Efficient Visual Grounding Language Models) training pipeline. It is built on top of Qwen3-VL-4B-Thinking.

EGM-Qwen3-VL-8B-SFT is the supervised fine-tuning (SFT) checkpoint from the first stage of the EGM (Efficient Visual Grounding Language Models) training pipeline. It is built on top of Qwen3-VL-8B-Thinking.

EGM-Qwen3-VL-4B is an efficient visual grounding model from the EGM (Efficient Visual Grounding Language Models) family. It is built on top of Qwen3-VL-4B-Thinking and trained with a two-stage pipeline: supervised fine-tuning (SFT) followed by reinforcement learning (RL) using GRPO (Group Relative Policy Optimization).

Poro 2 8B Math Reasoning SFT Preview is a specialized model focused on mathematical reasoning and problem-solving. This preview model was created through supervised fine-tuning of a context-extended base model. This model excels at mathematical reasoning tasks but is not optimized for general conversational use or other domains.

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

This model is a fine-tuned derivative of Qwen3.5-35B-A3B. Follow the Qwen3.5-35B-A3B serving guide for deployment with vLLM, replacing the model path with nvidia/NVIDIA-Ising-Calibration-1-35B-A3B. Suggested inference settings: temperature=0.2, maxtokens=16384.

harrier-oss-v1 is a family of multilingual text embedding models developed by Microsoft. The models use decoder-only architectures with last-token pooling and L2 normalization to produce dense text embeddings. They can be applied to a wide range of tasks, including but not limited to retrieval, clustering, semantic similarity, classification, bitext mining, and reranking. The models achieve state-of-the-art results on the Multilingual MTEB v2 benchmark as of the release date.

harrier-oss-v1 is a family of multilingual text embedding models developed by Microsoft. The models use decoder-only architectures with last-token pooling and L2 normalization to produce dense text embeddings. They can be applied to a wide range of tasks, including but not limited to retrieval, clustering, semantic similarity, classification, bitext mining, and reranking. The models achieve state-of-the-art results on the Multilingual MTEB v2 benchmark as of the release date.

We introduce UniRG-CXR, a radiology report generation model that obtains SOTA performance on ReXrank. More details can be found in the paper: Scaling medical imaging report generation with multimodal reinforcement learning

gemma 4 31B is een open-source taalmodel van Google met 31B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants. Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages.

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants. Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages.

Use temperature=1.0 and topp=0.95 across all tasks and serving backends — reasoning, tool calling, and general chat alike.

The pretraining data has a cutoff date of September 2024\.

Model Dates: Trained between Oct 2025 and March 2026

Model Summary: Granite-4.0-3B-Vision is a vision-language model (VLM) designed for enterprise-grade document data extraction. It focuses on specialized, complex extraction tasks that ultracompact models often struggle with:

EGM-Qwen3-VL-8B is the flagship model of the EGM (Efficient Visual Grounding Language Models) family. It is built on top of Qwen3-VL-8B-Thinking and trained with a two-stage pipeline: supervised fine-tuning (SFT) followed by reinforcement learning (RL) using GRPO (Group Relative Policy Optimization).

Qwen3.5 0.8B is een open-source taalmodel van Qwen met 0.8B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3.5 2B is een open-source taalmodel van Qwen met 2B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Model Summary: Granite-4.0-1b-speech is a compact and efficient speech-language model, specifically designed for multilingual automatic speech recognition (ASR) and bidirectional automatic speech translation (AST).

Qwen3.5 4B is een open-source taalmodel van Qwen met 4B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

[!Note] This repository contains model weights and configuration files for the pre-trained only model in the Hugging Face Transformers format. These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, etc. The intended use cases are fine-tuning, in-context learning experiments, and other research or development purposes, not direct interaction. However, the control tokens, e.g., <|imstart| and <|imend| were trained to allow efficient LoRA-style PEFT with the official chat template, mitigating the need to finetune embeddings, a significant optimization given Qwen3.5's larger

Qwen3.5 27B is een open-source taalmodel van Qwen met 27B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3.5 35B A3B is een open-source taalmodel van Qwen met 35B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

llama-nv-embed-reasoning-3b is a 3.2B-parameter embedding model designed to produce high‑quality sentence and document representations for retrieval, semantic search, and similarity tasks, with a strong focus on reasoning‑heavy content.

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

We introduce X-Reasoner, a vision-language model posttrained solely on general-domain text for generalizable reasoning, using a twostage approach: an initial supervised fine-tuning phase with distilled long chainof-thoughts, followed by reinforcement learning with verifiable rewards. Experiments show that X-Reasoner successfully transfers reasoning capabilities to both multimodal and out-of-domain settings, outperforming existing state-of-theart models trained with in-domain and multimodal data across various general and medical benchmarks. More details can be found in the paper: X-Reasoner: T

GLM-OCR is a multimodal OCR model for complex document understanding, built on the GLM-V encoder–decoder architecture. It introduces Multi-Token Prediction (MTP) loss and stable full-task reinforcement learning to improve training efficiency, recognition accuracy, and generalization. The model integrates the CogViT visual encoder pre-trained on large-scale image–text data, a lightweight cross-modal connector with efficient token downsampling, and a GLM-0.5B language decoder. Combined with a two-stage pipeline of layout analysis and parallel recognition based on PP-DocLayout-V3, GLM-OCR deliver

Chart2CSV is a specialized vision-language model fine-tuned for the accurate extraction of tabular data from charts and visualizations. Built on top of ibm-granite/granite-vision-3.3-2b, it produces machine-readable CSV outputs with improved numeric fidelity compared to general-purpose VLMs. The model is trained using code-guided synthetic chart data following the ChartGen methodology, which strengthens factual grounding and reduces hallucination in the Chart-to-CSV task.

The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni. Experiments show that the 1.7B version achieves state-of-the-art performance among open-source ASR models and is competitive with the strongest proprietary commercial APIs. Here are the main features:

The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni. Experiments show that the 1.7B version achieves state-of-the-art performance among open-source ASR models and is competitive with the strongest proprietary commercial APIs. Here are the main features:

The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni. Experiments show that the 1.7B version achieves state-of-the-art performance among open-source ASR models and is competitive with the strongest proprietary commercial APIs. Here are the main features:

[!NOTE] WARNING: Although this model has undergone safety and value alignment, it may still occasionally generate unintended or undesired outputs. Work In Progress New versions will be available during the coming weeks/months. Sampling Parameters: For optimal performance, we recommend using temperatures close to zero (0 - 0.2). Additionally, we advise against using any type of repetition penalty, as from our experience, it negatively impacts instructed model's responses.

Inference using Huggingface transformers on NVIDIA GPUs. Requirements tested on python 3.12.9 + CUDA11.8：

This is the model card for EuroLLM-9B-Instruct-2512, an improved version of utter-project/EuroLLM-9B-Instruct. In comparison with the previous version, this version includes the long-context extension phase and the revamped post-training recipe from utter-project/EuroLLM-22B-Instruct.

Developer: Microsoft Corporation Authorized Representative: Microsoft Ireland Operations Limited, 70 Sir John Rogerson's Quay, Dublin 2, D02 R296, Ireland Release Date: March 4, 2026 License: MIT Parameters: 15B Context Length: 16,384 tokens Inputs: Text and Images Outputs: Text Training GPUs: 240 B200s Training Time: 4 days Training Dates: February 3, 2025 – February 21, 2026 Model Dependencies: Phi-4-Reasoning

GLM-4.7-Flash is a 30B-A3B MoE model. As the strongest model in the 30B class, GLM-4.7-Flash offers a new option for lightweight deployment that balances performance and efficiency.

translategemma 27b it is een multimodaal taalmodel van Google met 29B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

translategemma 12b it is een multimodaal taalmodel van Google met 13B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

translategemma 4b it is een multimodaal taalmodel van Google met 5B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

GLM-Image is an image generation model adopts a hybrid autoregressive + diffusion decoder architecture. In general image generation quality, GLM‑Image aligns with mainstream latent diffusion approaches, but it shows significant advantages in text-rendering and knowledge‑intensive generation scenarios. It performs especially well in tasks requiring precise semantic understanding and complex information expression, while maintaining strong capabilities in high‑fidelity and fine‑grained detail generation. In addition to text‑to‑image generation, GLM‑Image also supports a rich set of image‑to‑imag

This model is a fine-tuned version of the openai/whisper-large-v3-turbo model finetuned for automatic speech recognition (ASR) in several Kenyan languages, including Swahili, Kalenjin, Kikuyu, Luo, Maasai and Somali. Whisper is a transformer-based encoder-decoder model that converts raw audio into text. The encoder processes audio inputs as log-Mel spectrograms, capturing acoustic and linguistic features, while the decoder generates text tokens in an autoregressive manner. This design allows the model to handle diverse languages, accents, and noise conditions with strong generalization.

medgemma 1.5 4b it is een multimodaal taalmodel van Google met 4.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Fine-tuning was performed on the entire unified multilingual ASR dataset, comprising the mentioned six languages, to encourage cross-lingual generalization. During fine-tuning, only the audio-specific components: audio embedding module, audio encoder, and audio projection layers, were unfrozen and set as trainable, while the rest of the model parameters remained frozen to preserve pretrained language capabilities. Dropout was applied to both the audio encoder and projection layers to regularize training. The model leverages a multimodal processor that handles text tokenization and audio featur

The Qwen3-VL-Embedding and Qwen3-VL-Reranker model series are the latest additions to the Qwen family, built upon the recently open-sourced and powerful Qwen3-VL foundation model. Specifically designed for multimodal information retrieval and cross-modal understanding, this suite accepts diverse inputs including text, images, screenshots, and videos, as well as inputs containing a mixture of these modalities.

The Qwen3-VL-Embedding and Qwen3-VL-Reranker model series are the latest additions to the Qwen family, built upon the recently open-sourced and powerful Qwen3-VL foundation model. Specifically designed for multimodal information retrieval and cross-modal understanding, this suite accepts diverse inputs including text, images, screenshots, and videos, as well as inputs containing a mixture of these modalities.

We are excited to introduce Qwen-Image-2512, the December update of Qwen-Image’s text-to-image foundational model. You are welcome to try the latest model at Qwen Chat. Compared to the base Qwen-Image model released in August, Qwen-Image-2512 features the following key improvements:

GLM-4.7, your new coding partner, is coming with the following features:

The post-training data has a cutoff date of November 28, 2025\. The pre-training data has a cutoff date of June 25, 2025\.

- Model Description - Intended Uses and Limitations - How to Get Started with the Model - Training Details - Citation - Additional Information

This is the model card for EuroMoE-2.6B-A0.6B-Instruct-2512. You can also check the pre-trained version: EuroMoE-2.6B-A0.6B-2512.

Cosmos Reason2 2B is een multimodaal taalmodel van NVIDIA met 2.4B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Cosmos Reason2 8B is een multimodaal taalmodel van NVIDIA met 8.8B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

- Model Description - Intended Uses and Limitations - How to Get Started with the Model - Training Details - Citation - Additional Information

GLM-ASR-Nano-2512 is a robust, open-source speech recognition model with 1.5B parameters. Designed for real-world complexity, it outperforms OpenAI Whisper V3 on multiple benchmarks while maintaining a compact size.

⚠️ This project is intended for research and educational purposes only. Any use for illegal data access, system interference, or unlawful activities is strictly prohibited. Please review our Terms of Use carefully.

⚠️ This project is intended for research and educational purposes only. Any use for illegal data access, system interference, or unlawful activities is strictly prohibited. Please review our Terms of Use carefully.

This model is part of the GLM-V family of models, introduced in the paper GLM-4.1V-Thinking and GLM-4.5V: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning.

This model is part of the GLM-V family of models, introduced in the paper GLM-4.1V-Thinking and GLM-4.5V: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning.

llama-nemotron-embed-vl-1b-v2 was developed by NVIDIA for multimodal question-answering retrieval. The model can embed document pages in the form of image, text, or combined image–text inputs. Documents can be retrieved given a user query in text form. The model supports page images containing text, tables, charts, and infographics. We report the evaluation of this model on two internal multimodal retrieval benchmarks, and on the popular ViDoRe V1 and V2 benchmarks and the new Vidore V3 benchmark.

This is the model card for EuroLLM-22B. You can also check the post-trained version: EuroLLM-22B-Instruct-2515.

[!Tip] This model was contributed by Xenova from Hugging Face. We sincerely appreciate the integration and community collaboration. While preliminary functionality checks have been performed, comprehensive testing has not yet been completed. We recommend you to proceed with caution and conducting your own evaluations for specific use cases. If any issues arise, open a PR/Issue here and we will try to address them promptly.

Salamandra-VL-7B-2512 is the latest version of the Salamandra vision model family. This version brings significant improvements in both architecture and training data.

Voxtral TTS is a frontier, open-weights text-to-speech model that’s fast, instantly adaptable, and produces lifelike speech for voice agents. The model is released with BF16 weights and a set of reference voices. These voices are licensed under CC BY-NC 4, which is the license that the model inherits.

- Repository: https://github.com/zheny2751-dotcom/WebVIA - Paper: https://arxiv.org/pdf/2511.06251

- Repository: https://github.com/zai-org/UI2CodeN - Paper: https://arxiv.org/abs/2511.08195

Kimi K2 Thinking is the latest, most capable version of open-source thinking model. Starting with Kimi K2, we built it as a thinking agent that reasons step-by-step while dynamically invoking tools. It sets a new state-of-the-art on Humanity's Last Exam (HLE), BrowseComp, and other benchmarks by dramatically scaling multi-step reasoning depth and maintaining stable tool-use across 200–300 sequential calls. At the same time, K2 Thinking is a native INT4 quantization model with 256k context window, achieving lossless reductions in inference latency and GPU memory usage.

Using the 🤗's Diffusers library to run URSA in a simple and efficient manner.

Using the 🤗's Diffusers library to run URSA in a simple and efficient manner.

Update: We just released Fara1.5 which improves on Fara-7B dramatically and is available in three model sizes 4B, 9B and 27B!

Kimi Linear is a hybrid linear attention architecture that outperforms traditional full attention methods across various contexts, including short, long, and reinforcement learning (RL) scaling regimes. At its core is Kimi Delta Attention (KDA)—a refined version of Gated DeltaNet that introduces a more efficient gating mechanism to optimize the use of finite-state RNN memory.

Kimi Linear is a hybrid linear attention architecture that outperforms traditional full attention methods across various contexts, including short, long, and reinforcement learning (RL) scaling regimes. At its core is Kimi Delta Attention (KDA)—a refined version of Gated DeltaNet that introduces a more efficient gating mechanism to optimize the use of finite-state RNN memory.

This model is for research and development only. <br

- Repository: https://github.com/thu-coai/Glyph - Paper: https://arxiv.org/abs/2510.17800

Using the 🤗's Diffusers library to run URSA in a simple and efficient manner.

Using the 🤗's Diffusers library to run URSA in a simple and efficient manner.

NVIDIA-Nemotron-Nano-VL-12B-V2-FP4-QAD is the quantized version of the NVIDIA Nemotron Nano VL V2 model, which is an auto-regressive vision language model that uses an optimized transformer architecture. For more information, please check here. The NVIDIA Nemotron Nano VL FP4 QAD model is quantized with TensorRT Model Optimizer.

Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

torch==2.6.0 transformers==4.46.3 tokenizers==0.20.3 einops addict easydict pip install flash-attn==2.7.3 --no-build-isolation

The Llama Nemotron Embedding 1B model is optimized for multilingual and cross-lingual text question-answering retrieval with support for long documents (up to 8192 tokens) and dynamic embedding size (Matryoshka Embeddings). This model was evaluated on 26 languages: English, Arabic, Bengali, Chinese, Czech, Danish, Dutch, Finnish, French, German, Hebrew, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Norwegian, Persian, Polish, Portuguese, Russian, Spanish, Swedish, Thai, and Turkish.

This model is for research and development only.

Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

functiongemma 270m it is een open-source taalmodel van Google met 0.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Model Summary: Granite-4.0-350M is a lightweight instruct model finetuned from Granite-4.0-350M-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques including supervised finetuning, reinforcement learning, and model merging.

Model Summary: Granite-4.0-1B-Base is a lightweight decoder-only language model designed for scenarios where efficiency and speed are critical. They can run on resource-constrained devices such as smartphones or IoT hardware, enabling offline and privacy-preserving applications. It also supports Fill-in-the-Middle (FIM) code completion through the use of specialized prefix and suffix tokens. The model is trained from scratch on approximately 15 trillion tokens following a four-stage training strategy: 10 trillion tokens in the first stage, 2 trillion in the second, another 2 trillion in the th

Model Summary: Granite-4.0-1B is a lightweight instruct model finetuned from Granite-4.0-1B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques including supervised finetuning, reinforcement learning, and model merging.

This model achieves state-of-the-art performance on the multilingual MTEB leaderboard as of October 21, 2025.

Llama-3.1-Nemotron-Nano-VL-8B-V1-FP4-QAD is the quantized version of the NVIDIA Llama Nemotron Nano VL model, which is an auto-regressive vision language model that uses an optimized transformer architecture. For more information, please check here. The NVIDIA Llama Nemotron Nano VL FP4 QAD model is quantized with TensorRT Model Optimizer.

Unlike typical LLMs that are trained to play the role of the "assistant" in conversation, we trained UserLM-8b to simulate the “user” role in conversation (by training it to predict user turns in a large corpus of conversations called WildChat). This model is useful in simulating more realistic conversations, which is in turn useful in the development of more robust assistants.

Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

Compared with GLM-4.5, GLM-4.6 brings several key improvements:

[!WARNING] WARNING: This is a language model that has undergone instruction tuning for conversational settings that exploit function calling capabilities. It has not been aligned with human preferences. As a result, it may generate outputs that are inappropriate, misleading, biased, or unsafe. These risks can be mitigated through additional post-training stages, which is strongly recommended before deployment in any production system, especially for high-stakes applications. How to use from datetime import datetime from transformers import AutoTokenizer, AutoModelForCausalLM import transformer

We are excited to announce the official release of DeepSeek-V3.2-Exp, an experimental version of our model. As an intermediate step toward our next-generation architecture, V3.2-Exp builds upon V3.1-Terminus by introducing DeepSeek Sparse Attention—a sparse attention mechanism designed to explore and validate optimizations for training and inference efficiency in long-context scenarios.

This update maintains the model's original capabilities while addressing issues reported by users, including:

Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.

gpt-oss-safeguard-120b and gpt-oss-safeguard-20b are safety reasoning models built-upon gpt-oss. With these models, you can classify text content based on safety policies that you provide and perform a suite of foundational safety tasks. These models are intended for safety use cases. For other applications, we recommend using gpt-oss models.

gpt-oss-safeguard-120b and gpt-oss-safeguard-20b are safety reasoning models built-upon gpt-oss. With these models, you can classify text content based on safety policies that you provide and perform a suite of foundational safety tasks. These models are intended for safety use cases. For other applications, we recommend using gpt-oss models.

📣 Update [10-07-2025]: Added a default system prompt to the chat template to guide the model towards more professional, accurate, and safe responses.

📣 Update [10-07-2025]: Added a default system prompt to the chat template to guide the model towards more professional, accurate, and safe responses.

📣 Update [10-07-2025]: Added a default system prompt to the chat template to guide the model towards more professional, accurate, and safe responses.

📣 Update [10-07-2025]: Added a default system prompt to the chat template to guide the model towards more professional, accurate, and safe responses.

Kimi K2-Instruct-0905 is the latest, most capable version of Kimi K2. It is a state-of-the-art mixture-of-experts (MoE) language model, featuring 32 billion activated parameters and a total of 1 trillion parameters.

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

embeddinggemma 300m qat q8 0 unquantized is een open-source taalmodel van Google met 0.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

embeddinggemma 300m qat q4 0 unquantized is een open-source taalmodel van Google met 0.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

DeepSeek-V3.1 is a hybrid model that supports both thinking mode and non-thinking mode. Compared to the previous version, this upgrade brings improvements in multiple aspects:

TowerVision is a family of open-source multilingual vision-language models with strong capabilities optimized for a variety of vision-language use cases, including image captioning, visual understanding, summarization, question answering, and more. TowerVision excels particularly in multimodal multilingual translation benchmarks and culturally-aware tasks, demonstrating exceptional performance across 20 languages and dialects.

DeepSeek-V3.1 is a hybrid model that supports both thinking mode and non-thinking mode. Compared to the previous version, this upgrade brings improvements in multiple aspects:

1. Model Summary 2. How to use 3. Evaluation 4. Training 5. Limitations 6. Legal Aspects

The pretraining data has a cutoff date of September 2024.

TowerVision is a family of open-source multilingual vision-language models with strong capabilities optimized for a variety of vision-language use cases, including image captioning, visual understanding, summarization, question answering, and more. TowerVision excels particularly in multimodal multilingual translation benchmarks and culturally-aware tasks, demonstrating exceptional performance across 20 languages and dialects.

This model is part of the GLM-V family of models, introduced in the paper GLM-4.1V-Thinking and GLM-4.5V: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning.

gemma 3 270m is een open-source taalmodel van Google met 0.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 4B Instruct 2507 is een open-source taalmodel van Qwen met 4B parameters en een contextvenster van 262K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Welcome to the gpt-oss series, OpenAI’s open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases.

Install the latest version of diffusers pip install git+https://github.com/huggingface/diffusers

Llama-3.3-Nemotron-Super-49B-v1.5-FP8 is a significantly upgraded version of Llama-3.3-Nemotron-Super-49B-v1 and is a large language model (LLM) which is a derivative of Meta Llama-3.3-70B-Instruct (AKA the reference model). It is a reasoning model that is post trained for reasoning, human chat preferences, and agentic tasks, such as RAG and tool calling. The model supports a context length of 128K tokens.

Qwen3-Coder is available in multiple sizes. Today, we're excited to introduce Qwen3-Coder-30B-A3B-Instruct. This streamlined model maintains impressive performance and efficiency, featuring the following key enhancements:

gemma 3 270m it is een open-source taalmodel van Google met 0.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

- Developed by: Fraunhofer, Forschungszentrum Jülich, TU Dresden, DFKI - Funded by: German Federal Ministry of Economics and Climate Protection (BMWK) in the context of the OpenGPT-X project - Model type: Transformer based decoder-only model - Language(s) (NLP): bg, cs, da, de, el, en, es, et, fi, fr, ga, hr, hu, it, lt, lv, mt, nl, pl, pt, ro, sk, sl, sv - Shared by: OpenGPT-X

We are excited to introduce Wan2.2, a major upgrade to our foundational video models. With Wan2.2, we have focused on incorporating the following innovations:

We are excited to introduce Wan2.2, a major upgrade to our foundational video models. With Wan2.2, we have focused on incorporating the following innovations:

Compared to nvidia/DeepSeek-R1-0528-FP4, this checkpoint additionally quantizes the wo module in attention layers.

The GLM-4.5 series models are foundation models designed for intelligent agents. GLM-4.5 has 355 billion total parameters with 32 billion active parameters, while GLM-4.5-Air adopts a more compact design with 106 billion total parameters and 12 billion active parameters. GLM-4.5 models unify reasoning, coding, and intelligent agent capabilities to meet the complex demands of intelligent agent applications.

The GLM-4.5 series models are foundation models designed for intelligent agents. GLM-4.5 has 355 billion total parameters with 32 billion active parameters, while GLM-4.5-Air adopts a more compact design with 106 billion total parameters and 12 billion active parameters. GLM-4.5 models unify reasoning, coding, and intelligent agent capabilities to meet the complex demands of intelligent agent applications.

The GLM-4.5 series models are foundation models designed for intelligent agents. GLM-4.5 has 355 billion total parameters with 32 billion active parameters, while GLM-4.5-Air adopts a more compact design with 106 billion total parameters and 12 billion active parameters. GLM-4.5 models unify reasoning, coding, and intelligent agent capabilities to meet the complex demands of intelligent agent applications.

The GLM-4.5 series models are foundation models designed for intelligent agents. GLM-4.5 has 355 billion total parameters with 32 billion active parameters, while GLM-4.5-Air adopts a more compact design with 106 billion total parameters and 12 billion active parameters. GLM-4.5 models unify reasoning, coding, and intelligent agent capabilities to meet the complex demands of intelligent agent applications.

We are excited to introduce Wan2.2, a major upgrade to our foundational video models. With Wan2.2, we have focused on incorporating the following innovations:

Model Summary: Granite-embedding-small-english-r2 is a 47M parameter dense biencoder embedding model from the Granite Embeddings collection that can be used to generate high quality text embeddings. This model produces embedding vectors of size 384 based on context length of upto 8192 tokens. Compared to most other open-source models, this model was only trained using open-source relevance-pair datasets with permissive, enterprise-friendly license, plus IBM collected and generated datasets.

Model Summary: Granite-embedding-english-r2 is a 149M parameter dense biencoder embedding model from the Granite Embeddings collection that can be used to generate high quality text embeddings. This model produces embedding vectors of size 768 based on context length of upto 8192 tokens. Compared to most other open-source models, this model was only trained using open-source relevance-pair datasets with permissive, enterprise-friendly license, plus IBM collected and generated datasets.

embeddinggemma 300m is een open-source taalmodel van Google met 0.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

- Model Description - Intended Uses and Limitations - How to Get Started with the Model - Training Details - Citation - Additional Information

- Model Description - Intended Uses and Limitations - How to Get Started with the Model - Training Details - Citation - Additional Information

- Model Description - Intended Uses and Limitations - How to Get Started with the Model - Training Details - Citation - Additional Information

The MediPhi Model Collection comprises 7 small language models of 3.8B parameters from the base model Phi-3.5-mini-instruct specialized in the medical and clinical domains. The collection is designed in a modular fashion. Five MediPhi experts are fine-tuned on various medical corpora (i.e. PubMed commercial, Medical Wikipedia, Medical Guidelines, Medical Coding, and open-source clinical documents) and merged back with the SLERP method in their base model to conserve general abilities. One model combined all five experts into one general expert with the multi-model merging method BreadCrumbs. F

Kimi K2 is a state-of-the-art mixture-of-experts (MoE) language model with 32 billion activated parameters and 1 trillion total parameters. Trained with the Muon optimizer, Kimi K2 achieves exceptional performance across frontier knowledge, reasoning, and coding tasks while being meticulously optimized for agentic capabilities.

medgemma 27b it is een multimodaal taalmodel van Google met 29B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

This model was converted to MLX format from ibm-granite/granite-docling-258M using mlx-vlm version 0.3.3. Refer to the original model card for more details on the model.

FLUX.1 Krea dev is een multimodaal taalmodel van Black-forest-labs met 12B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Dayhoff is an Atlas of both protein sequence data and generative language models — a centralized resource that brings together 3.34 billion protein sequences across 1.7 billion clusters of metagenomic and natural protein sequences (GigaRef), 46 million structure-derived synthetic sequences (BackboneRef), and 16 million multiple sequence alignments (OpenProteinSet). These models can natively predict zero-shot mutation effects on fitness, scaffold structural motifs by conditioning on evolutionary or structural context, and perform guided generation of novel proteins within specified families. Le

Kimi K2 is a state-of-the-art mixture-of-experts (MoE) language model with 32 billion activated parameters and 1 trillion total parameters. Trained with the Muon optimizer, Kimi K2 achieves exceptional performance across frontier knowledge, reasoning, and coding tasks while being meticulously optimized for agentic capabilities.

Vision-Language Models (VLMs) have become foundational components of intelligent systems. As real-world AI tasks grow increasingly complex, VLMs must evolve beyond basic multimodal perception to enhance their reasoning capabilities in complex tasks. This involves improving accuracy, comprehensiveness, and intelligence, enabling applications such as complex problem solving, long-context understanding, and multimodal agents.

Vision-Language Models (VLMs) have become foundational components of intelligent systems. As real-world AI tasks grow increasingly complex, VLMs must evolve beyond basic multimodal perception to enhance their reasoning capabilities in complex tasks. This involves improving accuracy, comprehensiveness, and intelligence, enabling applications such as complex problem solving, long-context understanding, and multimodal agents.

Cosmos Predict2 0.6B Text2Image is een multimodaal taalmodel van NVIDIA met 0.6B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Phi-tiny-MoE is a lightweight Mixture of Experts (MoE) model with 3.8B total parameters and 1.1B activated parameters. It is compressed and distilled from the base model shared by Phi-3.5-MoE and GRIN-MoE using the SlimMoE approach, then post-trained via supervised fine-tuning and direct preference optimization for instruction following and safety. The model is trained on Phi-3 synthetic data and filtered public documents, with a focus on high-quality, reasoning-dense content. It is part of the SlimMoE series, which includes a larger variant, Phi-mini-MoE, with 7.6B total and 2.4B activated pa

Phi-mini-MoE is a lightweight Mixture of Experts (MoE) model with 7.6B total parameters and 2.4B activated parameters. It is compressed and distilled from the base model shared by Phi-3.5-MoE and GRIN-MoE using the SlimMoE approach, then post-trained via supervised fine-tuning and direct preference optimization for instruction following and safety. The model is trained on Phi-3 synthetic data and filtered public documents, with a focus on high-quality, reasoning-dense content. It is part of the SlimMoE series, which includes a smaller variant, Phi-tiny-MoE, with 3.8B total and 1.1B activated p

[!Note] This is an improved version of Kimi-VL-A3B-Thinking. Please consider using this updated model instead of the previous version.

Phi-4-mini-flash-reasoning is a lightweight open model built upon synthetic data with a focus on high-quality, reasoning dense data further finetuned for more advanced math reasoning capabilities. The model belongs to the Phi-4 model family and supports 64K token context length.

We introduce Kimi-Dev-72B, our new open-source coding LLM for software engineering tasks. Kimi-Dev-72B achieves a new state-of-the-art on SWE-bench Verified among open-source models.

Note for most users: This is an intermediate checkpoint from our post-training pipeline. Most users should use Poro 2 70B Instruct instead, which includes an additional round of Direct Preference Optimization (DPO) for improved response quality and alignment. This SFT-only model is primarily intended for researchers interested in studying the effects of different post-training techniques.

Note for most users: This is an intermediate checkpoint from our post-training pipeline. Most users should use Poro 2 8B Instruct instead, which includes an additional round of Direct Preference Optimization (DPO) for improved response quality and alignment. This SFT-only model is primarily intended for researchers interested in studying the effects of different post-training techniques.

gemma 3n E2B it is een multimodaal taalmodel van Google met 5.4B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

⚠️ PREVIEW RELEASE: This is a preview version of EuroMoE-2.6B-A0.6B-Instruct-Preview. The model is still under development and may have limitations in performance and stability. Use with caution in production environments.

This is the model card for EuroLLM-2.6B-A0.6-2512, the pre-trained model for EuroLLM-2.6B-A0.6-2512-Instruct.

⚠️ PREVIEW RELEASE: This is a preview version of EuroVLM-1.7B. The model is still under development and may have limitations in performance and stability. Use with caution in production environments.

⚠️ PREVIEW RELEASE: This is a preview version of EuroVLM-9B. The model is still under development and may have limitations in performance and stability. Use with caution in production environments.

This is the model card for EuroLLM-9B-2512, an improved version of utter-project/EuroLLM-9B. In comparison with the previous version, this version includes the long-context extension phase from utter-project/EuroLLM-22B.

gemma 3n E4B it is een multimodaal taalmodel van Google met 7.8B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

The Qwen3 Embedding model series is the latest proprietary model of the Qwen family, specifically designed for text embedding and ranking tasks. Building upon the dense foundational models of the Qwen3 series, it provides a comprehensive range of text embeddings and reranking models in various sizes (0.6B, 4B, and 8B). This series inherits the exceptional multilingual capabilities, long-text understanding, and reasoning skills of its foundational model. The Qwen3 Embedding series represents significant advancements in multiple text embedding and ranking tasks, including text retrieval, code re

The Qwen3 Embedding model series is the latest proprietary model of the Qwen family, specifically designed for text embedding and ranking tasks. Building upon the dense foundational models of the Qwen3 series, it provides a comprehensive range of text embeddings and reranking models in various sizes (0.6B, 4B, and 8B). This series inherits the exceptional multilingual capabilities, long-text understanding, and reasoning skills of its foundational model. The Qwen3 Embedding series represents significant advancements in multiple text embedding and ranking tasks, including text retrieval, code re

This model was introduced in the paper GUI-Actor: Coordinate-Free Visual Grounding for GUI Agents. It is developed based on UI-TARS-2B-SFT and is designed to predict the correctness of an action position given a language instruction. This model is well-suited for GUI-Actor, as its attention map effectively provides diverse candidates for verification with only a single inference.

DeepSeek R1 0528 Qwen3 8B is een open-source taalmodel van DeepSeek met 8B parameters en een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Poro 2 70B Base is a 70B parameter decoder-only transformer created through continued pretraining of Llama 3.1 70B to add Finnish language capabilities. It was trained on 165B tokens using a carefully balanced mix of Finnish, English, code, and math data. Poro 2 is a fully open source model and is made available under the Llama 3.1 Community License.

Poro 2 8B Base is an 8B parameter decoder-only transformer created through continued pretraining of Llama 3.1 8B to add Finnish language capabilities. It was trained on 165B tokens using a carefully balanced mix of Finnish, English, code, and math data. Poro 2 is a fully open source model and is made available under the Llama 3.1 Community License.

Poro 2 70B Instruct is an instruction-following chatbot model created through supervised fine-tuning (SFT) and Direct Preference Optimization (DPO) of the Poro 2 70B Base model. This model is designed for conversational AI applications and instruction following in both Finnish and English. It was trained on a carefully curated mix of English and Finnish instruction data, followed by preference tuning to improve response quality.

Poro 2 8B Instruct is an instruction-following chatbot model created through supervised fine-tuning (SFT) and Direct Preference Optimization (DPO) of the Poro 2 8B Base model. This model is designed for conversational AI applications and instruction following in both Finnish and English. It was trained on a carefully curated mix of English and Finnish instruction data, followed by preference tuning to improve response quality.

[!WARNING] WARNING: This model has been deprecated and is no longer recommended. For the latest model, please visit: https://huggingface.co/BSC-LT/Salamandra-VL-7B-2512

medgemma 4b it is een multimodaal taalmodel van Google met 4.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Granite Docling 258M builds upon the Idefics3 architecture, but introduces two key modifications: it replaces the vision encoder with siglip2-base-patch16-512 and substitutes the language model with a Granite 165M LLM. Try out our Granite-Docling-258 demo today.

2025-4-2 🌟🌟 BGE-VL models are also available on WiseModel.

2025-04-06 🚀🚀 MVRB Dataset are released on Huggingface: MVRB

For more details please refer to our Github: FlagEmbedding.

2025-4-2 🌟🌟 BGE-VL models are also available on WiseModel.

- Model Description - Intended Uses and Limitations - How to Get Started with the Model - Training Details - Citation - Additional Information

- Model Description - Intended Uses and Limitations - How to Get Started with the Model - Training Details - Citation - Additional Information

NextCoder: Robust Adaptation of Code LMs to Diverse Code Edits (ICML'2025)

Model Summary: Granite-4-Tiny-Preview is a 7B parameter fine-grained hybrid mixture-of-experts (MoE) instruct model fine-tuned from Granite-4.0-Tiny-Base-Preview using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets tailored for solving long context problems. This model is developed using a diverse set of techniques with a structured chat format, including supervised fine-tuning, and model alignment using reinforcement learning.

Phi-4-mini-reasoning is a lightweight open model built upon synthetic data with a focus on high-quality, reasoning dense data further finetuned for more advanced math reasoning capabilities. The model belongs to the Phi-4 model family and supports 128K token context length.

Model Summary: Granite-speech-3.3-2b is a compact and efficient speech-language model, specifically designed for automatic speech recognition (ASR) and automatic speech translation (AST). Granite-speech-3.3-2b uses a two-pass design, unlike integrated models that combine speech and language into a single pass. Initial calls to granite-speech-3.3-2b will transcribe audio files into text. To process the transcribed text using the underlying Granite language model, users must make a second call as each step must be explicitly initiated.

Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support, with the following key features:

Qwen3 32B is een open-source taalmodel van Qwen met 32B parameters en een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 30B A3B is een open-source taalmodel van Qwen met 30B parameters en een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 14B is een open-source taalmodel van Qwen met 14B parameters en een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 8B is een open-source taalmodel van Qwen met 8B parameters en een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support, with the following key features:

Qwen3 1.7B is een open-source taalmodel van Qwen met 1.7B parameters en een contextvenster van 41K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support, with the following key features:

We present Kimi-Audio, an open-source audio foundation model excelling in audio understanding, generation, and conversation. This repository hosts the model checkpoints for Kimi-Audio-7B.

We present Kimi-Audio, an open-source audio foundation model excelling in audio understanding, generation, and conversation. This repository hosts the model checkpoints for Kimi-Audio-7B-Instruct.

Llama Guard 4 12B is een multimodaal taalmodel van Meta met 12B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Cosmos Predict2 14B Text2Image is een multimodaal taalmodel van NVIDIA met 14B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Cosmos Predict2 2B Text2Image is een multimodaal taalmodel van NVIDIA met 2B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

NVIDIA Cosmos Reason – an open, customizable, 7B-parameter reasoning vision language model (VLM) for physical AI and robotics - enables robots and vision AI agents to reason like humans, using prior knowledge, physics understanding and common sense to understand and act in the real world. This model understands space, time, and fundamental physics, and can serve as a planning model to reason what steps an embodied agent might take next.

[!IMPORTANT] To fully take advantage of the model's capabilities, inference must use temperature=0.8, topk=50, topp=0.95, and dosample=True. For more complex queries, set maxnewtokens=32768 to allow for longer chain-of-thought (CoT).

Model Summary: Granite-speech-3.3-8b is a compact and efficient speech-language model, specifically designed for automatic speech recognition (ASR) and automatic speech translation (AST). Granite-speech-3.3-8b uses a two-pass design, unlike integrated models that combine speech and language into a single pass. Initial calls to granite-speech-3.3-8b will transcribe audio files into text. To process the transcribed text using the underlying Granite language model, users must make a second call as each step must be explicitly initiated.

[!IMPORTANT] To fully take advantage of the model's capabilities, inference must use temperature=0.8, topk=50, topp=0.95, and dosample=True. For more complex queries, set maxnewtokens=32768 to allow for longer chain-of-thought (CoT).

Model Summary: Granite-3.3-8B-Instruct is a 8-billion parameter 128K context length language model fine-tuned for improved reasoning and instruction-following capabilities. Built on top of Granite-3.3-8B-Base, the model delivers significant gains on benchmarks for measuring generic performance including AlpacaEval-2.0 and Arena-Hard, and improvements in mathematics, coding, and instruction following. It supports structured reasoning through \<think\\<\/think\ and \<response\\<\/response\ tags, providing clear separation between internal thoughts and final outputs. The model has been trained on

Model Summary: Granite-3.3-2B-Instruct is a 2-billion parameter 128K context length language model fine-tuned for improved reasoning and instruction-following capabilities. Built on top of Granite-3.3-2B-Base, the model delivers significant gains on benchmarks for measuring generic performance including AlpacaEval-2.0 and Arena-Hard, and improvements in mathematics, coding, and instruction following. It supports structured reasoning through \<think\\<\/think\ and \<response\\<\/response\ tags, providing clear separation between internal thoughts and final outputs. The model has been trained on

[!Warning] This model has a new version: Kimi-VL-A3B-Thinking-2506. Please consider using the new 2506 version for better abilties on general visual understanding, reasoning, video and agent scenarios.

We present Kimi-VL, an efficient open-source Mixture-of-Experts (MoE) vision-language model (VLM) that offers advanced multimodal reasoning, long-context understanding, and strong agent capabilities—all while activating only 2.8B parameters in its language decoder (Kimi-VL-A3B).

gemma 3 12b it qat q4 0 unquantized is een multimodaal taalmodel van Google met 12B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

The GLM family welcomes a new generation of open-source models, the GLM-4-32B-0414 series, featuring 32 billion parameters. Its performance is comparable to OpenAI's GPT series and DeepSeek's V3/R1 series, and it supports very user-friendly local deployment features. GLM-4-32B-Base-0414 was pre-trained on 15T of high-quality data, including a large amount of reasoning-type synthetic data, laying the foundation for subsequent reinforcement learning extensions. In the post-training stage, in addition to human preference alignment for dialogue scenarios, we also enhanced the model's performance i

The GLM family welcomes a new generation of open-source models, the GLM-4-32B-0414 series, featuring 32 billion parameters. Its performance is comparable to OpenAI's GPT series and DeepSeek's V3/R1 series, and it supports very user-friendly local deployment features. GLM-4-32B-Base-0414 was pre-trained on 15T of high-quality data, including a large amount of reasoning-type synthetic data, laying the foundation for subsequent reinforcement learning extensions. In the post-training stage, in addition to human preference alignment for dialogue scenarios, we also enhanced the model's performance i

The GLM family welcomes new members, the GLM-4-32B-0414 series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series. It also supports very user-friendly local deployment features. GLM-4-32B-Base-0414 was pre-trained on 15T of high-quality data, including substantial reasoning-type synthetic data. This lays the foundation for subsequent reinforcement learning extensions. In the post-training stage, we employed human preference alignment for dialogue scenarios. Additionally, using techniques like rejection sampling and reinforc

The GLM family welcomes new members, the GLM-4-32B-0414 series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series. It also supports very user-friendly local deployment features. GLM-4-32B-Base-0414 was pre-trained on 15T of high-quality data, including substantial reasoning-type synthetic data. This lays the foundation for subsequent reinforcement learning extensions. In the post-training stage, we employed human preference alignment for dialogue scenarios. Additionally, using techniques like rejection sampling and reinforc

The GLM family welcomes new members, the GLM-4-32B-0414 series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series. It also supports very user-friendly local deployment features. GLM-4-32B-Base-0414 was pre-trained on 15T of high-quality data, including substantial reasoning-type synthetic data. This lays the foundation for subsequent reinforcement learning extensions. In the post-training stage, we employed human preference alignment for dialogue scenarios. Additionally, using techniques like rejection sampling and reinforc

Llama 4 Maverick 17B 128E is een open-source taalmodel van Meta met 17B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Llama 4 Scout 17B 16E is een multimodaal taalmodel van Meta met 109B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Llama 4 Scout 17B 16E Instruct is een multimodaal taalmodel van Meta met 109B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Llama 4 Maverick 17B 128E Instruct is een open-source taalmodel van Meta met 17B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Model Summary: Granite-speech-3.2-8b is a compact and efficient speech-language model, specifically designed for automatic speech recognition (ASR) and automatic speech translation (AST). Granite-speech-3.2-8b uses a two-pass design, unlike integrated models that combine speech and language into a single pass. Initial calls to granite-speech-3.2-8b will transcribe audio files into text. To process the transcribed text using the underlying Granite language model, users must make a second call as each step must be explicitly initiated.

DeepSeek-V3-0324 demonstrates notable improvements over its predecessor, DeepSeek-V3, in several key aspects.

txgemma 2b predict is een open-source taalmodel van Google met 2.6B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

In the past five months since Qwen2-VL’s release, numerous developers have built new models on the Qwen2-VL vision-language models, providing us with valuable feedback. During this period, we focused on building more useful vision-language models. Today, we are excited to introduce the latest addition to the Qwen family: Qwen2.5-VL.

salamandra 7b instruct tools is een open-source taalmodel van BSC-LT met 7.8B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Llama-3.1-Nemotron-Nano-8B-v1 is a large language model (LLM) which is a derivative of Meta Llama-3.1-8B-Instruct (AKA the reference model). It is a reasoning model that is post trained for reasoning, human chat preferences, and tasks, such as RAG and tool calling.

Llama-3.3-Nemotron-Super-49B-v1 is a large language model (LLM) which is a derivative of Meta Llama-3.3-70B-Instruct (AKA the reference model). It is a reasoning model that is post trained for reasoning, human chat preferences, and tasks, such as RAG and tool calling. The model supports a context length of 128K tokens.

gemma 3 1b it is een open-source taalmodel van Google met 1B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

2024-12-27 🚀🚀 BGE-VL-CLIP models are released on Huggingface: BGE-VL-base and BGE-VL-large.

2024-12-27 🚀🚀 BGE-VL-CLIP models are released on Huggingface: BGE-VL-base and BGE-VL-large.

+ Resolution: Width and height must be between 512px and 2048px, divisible by 32, and ensure the maximum number of pixels does not exceed 2^21 px. + Precision: BF16 / FP32 (FP16 is not supported as it will cause overflow resulting in completely black images)

gemma 3 12b it is een multimodaal taalmodel van Google met 12B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

gemma 3 12b pt is een multimodaal taalmodel van Google met 12B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

gemma 3 27b it is een multimodaal taalmodel van Google met 27B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

gemma 3 27b pt is een multimodaal taalmodel van Google met 27B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

In this repository, we present Wan2.1, a comprehensive and open suite of video foundation models that pushes the boundaries of video generation. Wan2.1 offers these key features: - 👍 SOTA Performance: Wan2.1 consistently outperforms existing open-source models and state-of-the-art commercial solutions across multiple benchmarks. - 👍 Supports Consumer-grade GPUs: The T2V-1.3B model requires only 8.19 GB VRAM, making it compatible with almost all consumer-grade GPUs. It can generate a 5-second 480P video on an RTX 4090 in about 4 minutes (without optimization techniques like quantization). Its p

In this repository, we present Wan2.1, a comprehensive and open suite of video foundation models that pushes the boundaries of video generation. Wan2.1 offers these key features: - 👍 SOTA Performance: Wan2.1 consistently outperforms existing open-source models and state-of-the-art commercial solutions across multiple benchmarks. - 👍 Supports Consumer-grade GPUs: The T2V-1.3B model requires only 8.19 GB VRAM, making it compatible with almost all consumer-grade GPUs. It can generate a 5-second 480P video on an RTX 4090 in about 4 minutes (without optimization techniques like quantization). Its p

In this repository, we present Wan2.1, a comprehensive and open suite of video foundation models that pushes the boundaries of video generation. Wan2.1 offers these key features: - 👍 SOTA Performance: Wan2.1 consistently outperforms existing open-source models and state-of-the-art commercial solutions across multiple benchmarks. - 👍 Supports Consumer-grade GPUs: The T2V-1.3B model requires only 8.19 GB VRAM, making it compatible with almost all consumer-grade GPUs. It can generate a 5-second 480P video on an RTX 4090 in about 4 minutes (without optimization techniques like quantization). Its p

In this repository, we present Wan2.1, a comprehensive and open suite of video foundation models that pushes the boundaries of video generation. Wan2.1 offers these key features: - 👍 SOTA Performance: Wan2.1 consistently outperforms existing open-source models and state-of-the-art commercial solutions across multiple benchmarks. - 👍 Supports Consumer-grade GPUs: The T2V-1.3B model requires only 8.19 GB VRAM, making it compatible with almost all consumer-grade GPUs. It can generate a 5-second 480P video on an RTX 4090 in about 4 minutes (without optimization techniques like quantization). Its p

2024-12-27 🚀🚀 BGE-VL-CLIP models are released on Huggingface: BGE-VL-base and BGE-VL-large.

2024-12-27 🚀🚀 BGE-VL-CLIP models are released on Huggingface: BGE-VL-base and BGE-VL-large.

🎉Phi-4: [mini-reasoning | reasoning] | [multimodal-instruct | onnx]; [mini-instruct | onnx]

- Weight Decay: Critical for scaling to larger models - Consistent RMS Updates: Enforcing a consistent root mean square on model updates

- Weight Decay: Critical for scaling to larger models - Consistent RMS Updates: Enforcing a consistent root mean square on model updates

gemma 3 1b pt is een open-source taalmodel van Google met 1B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

gemma 3 4b it is een multimodaal taalmodel van Google met 4.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

gemma 3 4b pt is een multimodaal taalmodel van Google met 4.3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

🎉Phi-4: [mini-reasoning | reasoning] | [multimodal-instruct | onnx]; [mini-instruct | onnx]

Model Summary: Granite-Embedding-30m-Sparse is a 30M parameter sparse biencoder embedding model from the Granite Experimental suite that can be used to generate high quality text embeddings. This model produces variable length bag-of-word like dictionary, containing expansions of sentence tokens and their corresponding weights and is trained using a combination of open source relevance-pair datasets with permissive, enterprise-friendly license, and IBM collected and generated datasets. While maintaining competitive scores on academic benchmarks such as BEIR, this model also performs well on ma

Model Summary: Granite-3.2-2B-Instruct is an 2-billion-parameter, long-context AI model fine-tuned for thinking capabilities. Built on top of Granite-3.1-2B-Instruct, it has been trained using a mix of permissively licensed open-source datasets and internally generated synthetic data designed for reasoning tasks. The model allows controllability of its thinking capability, ensuring it is applied only when required.

Model Summary: granite-vision-3.2-2b is a compact and efficient vision-language model, specifically designed for visual document understanding, enabling automated content extraction from tables, charts, infographics, plots, diagrams, and more. The model was trained on a meticulously curated instruction-following dataset, comprising diverse public datasets and synthetic datasets tailored to support a wide range of document understanding and general image tasks. It was trained by fine-tuning a Granite large language model with both image and text modalities.

Model Summary: granite-vision-3.1-2b-preview is a compact and efficient vision-language model, specifically designed for visual document understanding, enabling automated content extraction from tables, charts, infographics, plots, diagrams, and more. The model was trained on a meticulously curated instruction-following dataset, comprising diverse public datasets and synthetic datasets tailored to support a wide range of document understanding and general image tasks. It was trained by fine-tuning a Granite large language model (https://huggingface.co/ibm-granite/granite-3.1-2b-instruct) with

--- license: apache-2.0 language: - en pipelinetag: image-text-to-text tags: - multimodal libraryname: transformers ---

--- licensename: qwen-research licenselink: https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct/blob/main/LICENSE language: - en pipelinetag: image-text-to-text tags: - multimodal libraryname: transformers ---

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorpor

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorpor

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorpor

DeepSeek R1 Distill Llama 70B is een open-source taalmodel van DeepSeek met 70B parameters en een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

DeepSeek R1 Distill Llama 8B is een open-source taalmodel van DeepSeek met 8B parameters en een contextvenster van 131K tokens, gehost op Europese GPU's via een OpenAI-compatibele API.

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorpor

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorpor

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorpor

If you are using the weights from this repository, please update to

Cosmos 1.0 Diffusion 7B Text2World is een open-source taalmodel van NVIDIA met 7B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

More information please refer to our repo: https://github.com/VectorSpaceLab/OmniGen

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning

The CogAgent-9B-20241220 model is based on GLM-4V-9B, a bilingual open-source VLM base model. Through data collection and optimization, multi-stage training, and strategy improvements, CogAgent-9B-20241220 achieves significant advancements in GUI perception, inference prediction accuracy, action space completeness, and task generalizability. The model supports bilingual (Chinese and English) interaction with both screenshots and language input.

Granite Guardian 3.1 2B is a fine-tuned Granite 3.1 2B Instruct model designed to detect risks in prompts and responses. It can help with risk detection along many key dimensions catalogued in the IBM AI Risk Atlas. It is trained on unique data comprising human annotations and synthetic data informed by internal red-teaming. It outperforms other open-source models in the same space on standard benchmarks.

Using the 🤗's Diffusers library to run NOVA in a simple and efficient manner.

Using the 🤗's Diffusers library to run NOVA in a simple and efficient manner.

Using the 🤗's Diffusers library to run NOVA in a simple and efficient manner.

Using the 🤗's Diffusers library to run NOVA in a simple and efficient manner.

Using the 🤗's Diffusers library to run NOVA in a simple and efficient manner.

Our training data is an extension of the data used for Phi-3 and includes a wide variety of sources from:

- Developed by: Fraunhofer, Forschungszentrum Jülich, TU Dresden, DFKI - Funded by: German Federal Ministry of Economics and Climate Protection (BMWK) in the context of the OpenGPT-X project - Model type: Transformer based decoder-only model - Language(s) (NLP): bg, cs, da, de, el, en, es, et, fi, fr, ga, hr, hu, it, lt, lv, mt, nl, pl, pt, ro, sk, sl, sv - Shared by: OpenGPT-X

[!WARNING] WARNING: This is a base language model that has not undergone instruction tuning or alignment with human preferences. As a result, it may generate outputs that are inappropriate, misleading, biased, or unsafe. These risks can be mitigated through additional post-training stages, which is strongly recommended before deployment in any production system, especially for high-stakes applications.

Model Summary: Granite-3.1-1B-A400M-Instruct is a 1B parameter long-context instruct model finetuned from Granite-3.1-1B-A400M-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets tailored for solving long context problems. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging.

Model Summary: Granite-3.1-2B-Instruct is a 2B parameter long-context instruct model finetuned from Granite-3.1-2B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets tailored for solving long context problems. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging.

Model Summary: Granite-3.1-8B-Instruct is a 8B parameter long-context instruct model finetuned from Granite-3.1-8B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets tailored for solving long context problems. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging.

Model Summary: Granite-Embedding-278M-Multilingual is a 278M parameter model from the Granite Embeddings suite that can be used to generate high quality text embeddings. This model produces embedding vectors of size 768 and is trained using a combination of open source relevance-pair datasets with permissive, enterprise-friendly license, and IBM collected and generated datasets. This model is developed using contrastive finetuning, knowledge distillation and model merging for improved performance.

Model Summary: Granite-Embedding-107M-Multilingual is a 107M parameter dense biencoder embedding model from the Granite Embeddings suite that can be used to generate high quality text embeddings. This model produces embedding vectors of size 384 and is trained using a combination of open source relevance-pair datasets with permissive, enterprise-friendly license, and IBM collected and generated datasets. This model is developed using contrastive finetuning, knowledge distillation and model merging for improved performance.

Model Summary: Granite-Embedding-30m-English is a 30M parameter dense bi-encoder embedding model from the Granite Embeddings suite that can be used to generate high quality text embeddings. This model produces embedding vectors of size 384 and is trained using a combination of open source relevance-pair datasets with permissive, enterprise-friendly license, and IBM collected and generated datasets. While maintaining competitive scores on academic benchmarks such as BEIR, this model also performs well on many enterprise use cases. This model is developed using retrieval oriented pre-training, c

News: Granite Embedding R2 models with 8192 context length released.

This repository provides the Depth ControlNet for Stable Diffusion 3.5 Large..

This repository provides the Blur ControlNet for Stable Diffusion 3.5 Large..

This repository provides the Canny ControlNet for Stable Diffusion 3.5 Large..

Install the transformers library from the source code:

Install the transformers library from the source code:

EuroLLM 9B Instruct is een open-source taalmodel van Utter-project met 9.2B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

EuroLLM 9B is een open-source taalmodel van Utter-project met 9.2B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

paligemma2 3b pt 896 is een multimodaal taalmodel van Google met 3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

paligemma2 3b pt 448 is een multimodaal taalmodel van Google met 3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

paligemma2 3b pt 224 is een multimodaal taalmodel van Google met 3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

paligemma2 3b ft docci 448 is een multimodaal taalmodel van Google met 3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

paligemma2 3b mix 224 is een multimodaal taalmodel van Google met 3B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

FLUX.1 Depth dev is een multimodaal taalmodel van Black-forest-labs met 12B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

FLUX.1 Canny dev is een multimodaal taalmodel van Black-forest-labs met 12B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

Install the transformers library from the source code:

Install the transformers library from the source code:

Salamandra is a highly multilingual model pre-trained from scratch that comes in three different sizes — 2B, 7B and 40B parameters — with their respective base and instruction-tuned variants. This model card corresponds to the 7B instructed version specific for AinaHack, an event launched by Generalitat de Catalunya to create AI tools for the Catalan administration.

Salamandra is a highly multilingual model pre-trained from scratch that comes in three different sizes — 2B, 7B and 40B parameters — with their respective base and instruction-tuned variants. This model card corresponds to the 2B instructed version specific for AinaHack, an event launched by Generalitat de Catalunya to create AI tools for the Catalan administration.

Qwen2.5-Coder is the latest series of Code-Specific Qwen large language models (formerly known as CodeQwen). As of now, Qwen2.5-Coder has covered six mainstream model sizes, 0.5, 1.5, 3, 7, 14, 32 billion parameters, to meet the needs of different developers. Qwen2.5-Coder brings the following improvements upon CodeQwen1.5:

Qwen2.5-Coder is the latest series of Code-Specific Qwen large language models (formerly known as CodeQwen). As of now, Qwen2.5-Coder has covered six mainstream model sizes, 0.5, 1.5, 3, 7, 14, 32 billion parameters, to meet the needs of different developers. Qwen2.5-Coder brings the following improvements upon CodeQwen1.5:

CogVideoX is an open-source video generation model similar to QingYing. The table below displays the list of video generation models we currently offer, along with their foundational information.

This model is the gptq-quantized version of Salamandra-2b for speculative decoding.

This model is the gptq-quantized version of Salamandra-7b for speculative decoding.

stable diffusion 3.5 medium is een multimodaal taalmodel van Stabilityai met 2.5B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

- Developed by: Fraunhofer, Forschungszentrum Jülich, TU Dresden, DFKI - Funded by: German Federal Ministry of Economics and Climate Protection (BMWK) in the context of the OpenGPT-X project - Model type: Transformer based decoder-only model - Language(s) (NLP): bg, cs, da, de, el, en, es, et, fi, fr, ga, hr, hu, it, lt, lv, mt, nl, pl, pt, ro, sk, sl, sv - Shared by: OpenGPT-X

Below is the model card of Emu3-Chat model, which is adapted from the original Emu3 model card that you can find here.

Below is the model card of Emu3-Chat model, which is adapted from the original Emu3 model card that you can find here.

If you are using the weights from this repository, please update to

If you are using the weights from this repository, please update to

stable diffusion 3.5 large turbo is een multimodaal taalmodel van Stabilityai met 8.1B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

stable diffusion 3.5 large is een multimodaal taalmodel van Stabilityai met 8.1B parameters, gehost op Europese GPU's via een OpenAI-compatibele API.

This model hub includes a finetuned version of YOLOv8 and a finetuned BLIP-2 model on the above dataset respectively. For more details of the models used and finetuning, please refer to the paper.

This model is the DiT version of CogView3, a text-to-image generation model, supporting image generation from 512 to 2048px.

Model Summary: Granite-3.0-1B-A400M-Instruct is an 1B parameter model finetuned from Granite-3.0-1B-A400M-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging.