Your localrouting layer
One layer for every model you can reach — API, local, and subscription. Start free.
No card required
Works with the tools you already use. Routes to every major provider — and the models on your own machine
Included inference.Your first request works with nothing connected.
Every account comes with inference included. Download, run your tool, and it answers. No keys, no card.
Memory and integrations,in every agent.
Shared memory across your agents, Context.dev included, plus any integration from the marketplace.
Re-explaining your project?
Memory carries it from one agent to the next. Free on every plan.
Web data with
Context.dev
Any page as Markdown, JSON or a screenshot. Included.
Setting up tools in every client?
Add it once from the marketplace. It shows up everywhere.
Running in two steps.
Run your tools
Inference is included — your first request works with nothing connected.
- 1# install once
- 2curl -fsSL https://www.openllm.sh/install | bash
- 3
- 4# then run your tool through OpenLLM
- 5openllm claude
- 6
- 7# your ~/.claude config is never modified
The problems you stop having.
Rate-limited mid-task?
The next model in your chain answers. Your agent never stops.
Keys scattered across tools?
Every subscription, local model, and API key you have, in every tool.
Models stuck on one machine?
Your devices and the API act as one network.
Only your combination opens it.
Provider keys are AES-256-GCM encrypted in your browser, under a recovery phrase only you hold. We store ciphertext. The key to it rides inside your sk-llm key, lives in memory for one request, then is zeroed.
- 01
No master key exists. We couldn't decrypt your vault if subpoenaed.
- 02
A breach leaks nothing usable. Our database holds ciphertext only.
- 03
Lose the phrase? Reset and re-paste. We can't recover it, by design.
The routing layerfor all inference.
API, local, subscription, free — every model, every device, one layer. Today for developers; tomorrow for every agent.
226
Models
In the catalog, ready to route
13
Providers
APIs, subscriptions, local servers
8
Tools
Claude Code, Codex, OpenCode and more
0
Keys to start
Inference is included
Free to start.
Every account starts with inference included. No card, no trial clock.
Free
$0
- Included inference
- Memory across your agents
- Context.dev web data
- 2 devices
Pro
More room
- Everything in Free
- 5 devices
- More included usage
- 3-day free trial