Skip to main content
These are the models the router inference endpoint can execute, with the identifier you pass as model (or pick in a routing config on the Routing page). Per-token prices for all of them live on the platform’s Pricing page (Inference → Pricing); prices are not repeated here so they can’t go stale.

​
Identifier spelling

Belvedir uses provider-prefixed identifiers for every provider except OpenAI, whose bare gpt-* ids are the convention. Bare ids from provider-native SDKs are normalized server-side (claude-sonnet-5 → anthropic/claude-sonnet-5, grok-4.6 → x-ai/grok-4.6, gemini-* → google/*, mistral-* → mistralai/*, deepseek-* → deepseek/*), so a call site migrated from another SDK keeps working with the id it already names.

​
Anthropic

Served directly on Anthropic’s API. That is every model Anthropic’s API serves; dated snapshot ids (claude-opus-4-5-20251101, …) resolve to the same rates.

​
OpenAI

Served directly on OpenAI’s API. Bare ids are the norm; the openai/ prefix is also accepted. Any current OpenAI chat id (gpt-*, chatgpt-*, o-series) routes here; the priced line-up: That is every callable chat model OpenAI’s API serves, at OpenAI’s list prices; dated snapshot ids resolve to the same rates. OpenAI’s pro and codex models are served only on their Responses API, which the router doesn’t speak, and the versioned chat-latest ids are deprecated on OpenAI’s side; both return a clear error naming the base model to use instead.

​
xAI

Served through OpenRouter, priced from the provider’s reported cost per call.

​
Open models

Beyond this list, any OpenRouter model id works as a pass-through: the call executes on OpenRouter and bills at OpenRouter’s reported cost for that call. If your organization has Chinese models turned off under Project Permissions, Chinese-lab identifiers (Qwen, GLM, DeepSeek, Kimi) are unavailable for that project: the router never picks one, and a call naming one returns a clear error.

​
Batch tier

Models served on Anthropic, OpenAI, and Sail also have a deferred batch tier at half of list price. OpenRouter-served models have no batch tier.

​
Your own models

  • Local models (Ollama-style name:tag ids such as qwen3:4b, llama3.1:8b, gpt-oss:20b): the decision endpoint can route to them, but your app executes the call; the inference endpoint returns a clear 400 for them.
  • Fine-tuned models (your project’s trained models, by name): same rule; the decision endpoint routes to them and your app serves them.
  • Bring your own provider: register a model id with your own base URL and key under Cloud Inference in the dashboard, and the inference endpoint executes that id on your endpoint instead — metered but never billed. This lifts the 400 for local and fine-tuned ids too.