INTEGRATION
Replicate-style flow with Meridian
Drop-in replacement for the Replicate chat API. Every model Replicate hosts — plus models they don't — behind one endpoint with unified billing, automatic fallback, and zero cold starts.
Covered models
Meridian proxies every chat model available on Replicate, plus first-party models from OpenAI, Anthropic, Google, and Mistral — all through the same request format.
Llama 3.1 405B
Meta via Replicate
Mixtral 8x22B
Mistral via Replicate
Claude 3 Opus
Anthropic via Replicate
Gemma 2 27B
Google via Replicate
Command R+
Cohere via Replicate
DBRX 132B
Databricks via Replicate
+ every other model on Replicate's catalog — Meridian resolves them at runtime.