One API for every model
Chat Completions, the Anthropic Messages API and the Responses API, plus embeddings and transcription, all with the same key.
03 Serve $ hyai models
Every open model.
One key.
Shared per token, or running for you alone, from European data centres.
GLM
DeepSeek
Qwen
Kimi
Mistral
Llama
gpt-oss
Gemma
Loes
Nemotron
Claude Code
Codex
OpenCode
Cursor
Cline-
Loes
-
OpenAI SDK
Anthropic SDK
What it does
Shared per token,
or a GPU of your own.
The model garden lists the open models you can call right now. Shared models run per token on GPUs in Europe: you pay for what you use and there is no machine to manage. Every model has its own page with its context length, its price per million tokens and a quality score.
When a model has to work for you alone, run it on a dedicated GPU. You create and delete it yourself, in the dashboard or through the API, and you pay per GPU minute while it runs. A private connection can keep its port closed to the public internet.
One model name can have several sources behind it. The router sends each request to a source that is healthy and fast, keeps a conversation on the machine that holds its cache, and moves on when a machine slows down.
Read more
Serve/ Model garden
Why teams use it Built for
Built for
production traffic.
Cached tokens cost less
Repeated prompt prefixes are billed at a lower rate automatically. You send nothing extra.
A Bench score on every model
Every model carries a Bench score from public benchmarks, marked as measured or estimated, and the rankings show measured speed per model.
Your own model too
Serve a model from Hugging Face or your own finetune on a dedicated GPU, booted with the official vLLM recipe for that model.
Questions
Questions we get
Which models can I use?
Every model in the model garden that shows as available. We add new open models after we have tested them.
When is a dedicated GPU worth it?
When the work needs a model nobody else runs, a guaranteed speed or full isolation. For most teams shared per token stays cheaper until the volume gets very large.
Does it work with the OpenAI SDK?
Yes. Set the base URL to https://hostyourai.com/api/v1 and use your HostYourAI key. The Anthropic SDK works the same way.
The platform
One platform, four layers
Start with one layer and add the others when the work asks for it. They share one key, one balance and one Activity view.
Call your first
open model.
One key gives you every model in the garden. Top up credits and pay per token, without a subscription.
