LLM gateway
Prism
by RODMENA
One model name with several vendors behind it.
An OpenAI-compatible gateway for large language models. Clients use one model name with several vendors behind it. When a vendor runs out of budget, applies a rate limit or goes down, Prism fails over to the next one mid-request without interrupting the client. The API reference, tested against the live gateway with the client binaries themselves, documents anything that does not work.
- One model name
- Failover mid-request
- OpenAI-compatible
- Tested API reference
Edition 1.0.0. One download per request.