Launching today

Inferbase AI
The control plane for AI models: route, govern, observe
1 follower
The control plane for AI models: route, govern, observe
1 follower
One OpenAI-compatible API for every model. Send model: "auto" and each request is routed to the best model for the task on quality, cost and latency, with the reason kept on record. Assign routing policies, guardrails and rate limits to each key, use your own provider keys, cloud account or self-hosted endpoint, and fall back automatically when a provider fails.







Free Options
Launch Team

AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted
Hi Product Hunt.
We launched here in April as a model comparison tool. Since then it has become something different: a gateway that sits between your application and the models.
You point your OpenAI or Anthropic SDK at one API. With model set to "auto", each request is classified and sent to the model that fits it, and you can read why that model was chosen. If a provider fails, the request moves to the next route. Each key can carry its own routing policy, guardrails and limits, and you can run on our managed routes, your own provider keys, your cloud account or a self-hosted endpoint.
The catalog, comparison and GPU sizing tools from the first launch are still there and still free.
There is a free plan with no card. I'd like to hear where the routing picks the wrong model for you, and what you would need before putting it in front of production traffic.
We're also taking five design partner teams: inferbase.ai/design-partners