One key. Every model you already pay for.
Any endpoint routes through the ChatGPT, Claude, Grok, and Kimi subscriptions you already pay for. No per-token pricing, no markup. Works in your app, IDE, and CLI.
Get StartedBring the providers you already pay for — subscriptions, API keys, or your own endpoints.
What You Get
Bring your own subscriptions, API keys, and endpoints. OpenLLM turns them into one consistent, encrypted surface for every tool you use.
One Place for Every Subscription
Use the ChatGPT plan you already pay for. Plug in your Claude account, your Gemini login, your Kimi Code subscription. OpenLLM routes each request to the right place.
Live-Editable Fallback Chains
Order your chain: Claude first, GPT next, then Gemini. Down, rate-limited, or over context? OpenLLM slides to the next entry without dropping the request.
Unified Usage & Cost
Every request, across every provider, in one dashboard. See what each model costs you, how often it fails, and how your fallback chains behave in production.
Encrypted at Rest, by Your Seed Phrase
Provider keys are stored as ciphertext under a seed phrase only you hold. The unlock material rides your own API key, lives in memory just long enough to call upstream, then is wiped.
All Your LLM Needs in One Place
Stop juggling dashboards. Your AI life is split across five logins and three billing pages. OpenLLM collapses all of it into one signed-in surface that speaks the API your tools already expect.
Stop paying twice. You already pay for ChatGPT Plus, Claude Pro, Gemini. Let those subscriptions power your agents and your CLI — instead of burning metered API credits on the same models.
Smaller blast radius. In a normal proxy, every provider key sits in plaintext in the database. OpenLLM only ever stores ciphertext — reading a key requires a live request with your own API key, not a database snapshot.
Stop rewriting integrations. Every model, every modality — chat, embeddings, images, audio — behind the OpenAI and Anthropic APIs. If your tool already talks to them, it already talks to OpenLLM.
Common Questions
Everything you need to know about OpenLLM.
How are my provider keys stored?
As ciphertext only, encrypted under a seed phrase that only you hold. The unlock material rides your own API key, lives in memory just long enough to call the provider, then is wiped. A database dump reveals nothing.
Can I use the subscriptions I already pay for?
Yes. Connect your ChatGPT plan, Claude account, or Kimi Code subscription and let them power your agents, your IDE, and your CLI — no extra metered API credits for the same models.
Does OpenLLM mark up tokens?
No. OpenLLM doesn't resell tokens. Every call uses your own credentials — your provider, your account, your price.
Will it work with my existing tools?
If your tool speaks the OpenAI or Anthropic API, it speaks OpenLLM. Chat, embeddings, image generation, and audio all live behind the same unified /v1 endpoints.
What happens when a provider goes down or rate-limits?
Your fallback chain takes over. OpenLLM slides to the next model without dropping the request — mid-stream it hands off the partial response so the next model picks up where the last one left off.
Can I bring my own endpoint?
Yes. Paste a base URL and an API key — OpenRouter, a local Ollama host, any OpenAI- or Anthropic-compatible gateway — and its models show up everywhere built-in models do.
Have another question? Get in touch