Providers
The accounts you connect, and the two ways to connect them.
A provider is one AI account OpenLLM uses on your behalf. You connect providers once under Providers in the dashboard. There are two connection types, and the difference matters for both privacy and compliance.
API keys
Paste a provider API key and OpenLLM stores it encrypted. Supported today:
- Anthropic
- OpenAI
- Google Gemini
- Kimi
- Alibaba DashScope
- Z.AI
- MiniMax
- AWS Bedrock
These run through the gateway. On each request the gateway decrypts the key in memory, calls the provider, and wipes it. Nothing readable is stored at rest. See Security.
Subscription accounts
Connect a consumer subscription over OAuth instead of a metered API key:
- ChatGPT
- Gemini CLI
- Kimi Code
- Claude Code
Subscription accounts run through a small local daemon on your own machine, not the cloud. When a request needs one, it hops to your machine, runs against the official vendor CLI with its own credentials, and returns the result. Those credentials never route through the OpenLLM cloud. Installing the daemon is covered on Connect your tools, and why this design matters is on Compliance.
Mixing providers
You do not have to pick one. Connect several, then order them in a fallback chain so a tier tries your preferred account first and falls back to the next when it is unavailable. A common setup is a subscription first for cost, then a metered API key as the reliable backstop.