Model garden Op aanvraag

Kimi VL A3B Thinking 2506

Dit model draait als dedicated deployment op grote GPU's en staat niet standaard in de gedeelde playground. Neem contact op, dan zetten we het voor je klaar.

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

moonshotai/Kimi-VL-A3B-Thinking-2506 Op aanvraag
text+image->text · moonshotai · EU-hosted
16B
Parameters
Contextvenster
48GB
Minimale VRAM
POST /api/v1/chat/completions Op aanvraag

Specificaties

Parameters 16B
Minimale VRAM 48 GB
Architectuur KimiVLForConditionalGeneration (vLLM)
Licentie mit
Modaliteit text+image->text
Uitgebracht June 2025
Uitgever moonshotai ↗

Prijzen

€0.08
Input (per 1M tokens)
€0.25
Output (per 1M tokens)

Gedeelde EU-router, pay-per-token, scale-to-zero. Dedicated GPU-deployments worden per uur afgerekend, zie prijzen.

Niet beschikbaar op de gedeelde router. Prijs op aanvraag als dedicated GPU-deployment.

Direct aanroepen

Drop-in vervanger voor OpenAI: wijzig alleen de base-URL en de API-key. Ook het Anthropic-formaat (/v1/messages) wordt ondersteund.

curl https://hostyourai.com/api/v1/chat/completions \
  -H "Authorization: Bearer hyai-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "moonshotai/Kimi-VL-A3B-Thinking-2506",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Veelgestelde vragen

Kan ik Kimi VL A3B Thinking 2506 in de EU draaien?

Ja. HostYourAI draait Kimi VL A3B Thinking 2506 op GPU's in Europese datacenters via vLLM. Prompts en outputs verlaten de EU niet en er is geen Amerikaanse cloudprovider in de keten.

Is Kimi VL A3B Thinking 2506 hosten AVG/GDPR-compliant?

Ja. Alle verwerking vindt plaats binnen de EU, er is een verwerkersovereenkomst (DPA) beschikbaar en de subprocessor-lijst is openbaar. Open-source gewichten betekenen ook: geen training op jouw data.

Wat kost Kimi VL A3B Thinking 2506?

Via de gedeelde EU-router betaal je €0.08 per miljoen input-tokens en €0.25 per miljoen output-tokens, zonder vaste kosten. Voor hoge volumes of isolatie kun je Kimi VL A3B Thinking 2506 ook als dedicated GPU-instance per uur draaien.

Is de API compatibel met OpenAI?

Ja. Je gebruikt de standaard OpenAI-SDK's met een aangepaste base-URL (https://hostyourai.com/api/v1). Ook de Anthropic Messages API wordt ondersteund als drop-in.

Andere modellen van Moonshotai

Kimi K2.7 Code

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.

1059B Bekijk model →
Kimi K2.6

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.

1059B Bekijk model →
Kimi K2.5

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.

1059B Bekijk model →
Kimi K2 Thinking

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.

1058B 262K context Bekijk model →
Kimi Linear 48B A3B Base

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.

49B Bekijk model →
Kimi Linear 48B A3B Instruct

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.

49B Bekijk model →

Vraag toegang aan

Kimi VL A3B Thinking 2506 is nog niet standaard beschikbaar. Laat je gegevens achter, dan regelen we een dedicated deployment.

Vraag toegang aan