MITHUNAI Gateway · Live
One OpenAI-compatible endpoint for your model providers
MITHUNAI Gateway puts one OpenAI-compatible endpoint in front of the model providers you use. Provider keys live in an encrypted vault, requests are routed with failover between providers, and every key carries its own usage and budget for the traffic that goes through it. Point any client that speaks the OpenAI format at it.
Last updated
What MITHUNAI Gateway does
One endpoint, OpenAI-shaped
Clients that already speak the OpenAI API format call one address, whichever provider answers.
Keys in a vault, not in clients
Provider keys are stored encrypted in the gateway, so the applications that call it never hold them.
Routing, failover, usage and budgets
Requests are routed between providers with failover, and usage and budgets are kept per key for the traffic that goes through the gateway.
One model layer for every answer.
Every completion and embedding the answer engine makes goes through one model layer. It talks directly to Anthropic, OpenAI and Gemini with your deployment's credentials, and falls back to another configured provider when one fails. Chat sends its model calls through the MITHUNAI Gateway; Agents uses the provider credentials entered in its own board.
Usage and Cost, Normalised
The model layer normalises token usage across providers and estimates the cost of each call, so usage analytics read the same whichever model answered.
Streaming Answers
Completions stream from the model layer, and the API relays them to clients as Server-Sent Events.
Model Catalogue
Each assistant picks its model from the deployment catalogue, and a model that lacks a capability is refused before any call is made.
Provider Fallback
Direct adapters for Anthropic, OpenAI and Gemini, with retries, timeouts and fallback to another configured provider.
Credential Redaction & Rate Limits
Provider credentials are deployment-wide and redacted from errors and logs, and every API key has its own request budget.
How teams use it
1Add providers
A platform administrator adds the providers you use and their keys, once.
2Issue keys
Each application or team gets its own gateway key, with its own usage and budget.
3Call one endpoint
Clients send OpenAI-format requests to the gateway and never see a provider key.
Who it is for
Platform teams that manage model providers for several applications and want one place for keys, routing and spend.
Read the documentation for this product or open it at gateway.mithunai.com.
Where it stands today
- MITHUNAI Chat sends its model calls through the gateway; the RAG answer engine calls providers with its own credentials, so its traffic is not counted here.
- It admits platform administrators, and an organisation’s owners and administrators where a deployment allows it.
Questions people ask first
The other MITHUNAI products, on the same account
- MITHUNAI RAG
Live
Cited answers from your documentation, on your site, over an API and over MCP.
- MITHUNAI Chat
Live
A chat workspace for your team, with cited answers from your own knowledge.
- MITHUNAI AI search visibility
Beta. Not yet run against a real AI search engine.
How AI search engines answer your buyers’ questions, and whom they cite.
- MITHUNAI Team
Beta. Access on request.
Employees for support, IT, DevOps, marketing, sales and finance that act only with approval.
Grounded answers from your own knowledge, a chat workspace for your team and one model gateway, on one account.
A 30-minute walkthrough on your own use case: connect a source, ask it questions, check the citations and see what it does when the answer is not there. Or ask the public playground now, with no account.