Supported models

Unified access to major providers through one API — use any model ID your workspace allows.

You can request any model ID your workspace's allowlist permits. There's no separate "Relixr model catalog" to learn — you use each provider's own model names.

ProviderExample model IDs
OpenAIgpt-4o, gpt-4o-mini, text-embedding-3-small
Anthropicclaude-sonnet-4-5, claude-haiku-4-5
Googlegemini-2.0-flash, gemini-2.5-pro
Mistralmistral-large-latest
Groqopenai/gpt-oss-20b (free tier), openai/gpt-oss-120b
Tip:

Not sure which model to pick? Start with a small, fast model like gpt-4o-mini or claude-haiku-4-5 for prototyping, then add a routing rule to upgrade to a larger model only for the requests that need it.

Model pricing is passed through at provider cost plus a 5% platform fee. Current rates live on the Models page so they stay accurate as providers change their pricing.

Next steps
Model routing
Route by cost, latency, or as a fallback chain.
Billing & credits
How the platform fee and provider cost are billed.
Make a requestChat completions