Use LibertAI in your tools
LibertAI is an OpenAI-compatible API at the base URL https://api.libertai.io/v1. Any tool that lets you set a custom OpenAI-compatible endpoint works with it: coding agents, IDE extensions, frameworks, desktop apps. You keep your prompts private, run open-weights models, and can leave any time, since nothing in your setup is proprietary to us.
| Tool | Type | Guide |
|---|---|---|
| Cline | VS Code coding agent | Set up Cline |
| Roo Code | VS Code coding agent | Set up Roo Code |
| Continue | IDE assistant (VS Code, JetBrains) | Set up Continue |
| OpenCode | Terminal coding agent | Set up OpenCode |
| Aider | Terminal pair programmer | Set up Aider |
| Claude Code | Terminal coding agent | Set up Claude Code |
| LangChain & LlamaIndex | Python frameworks | Set up LangChain & LlamaIndex |
Using something else? If it has an "OpenAI compatible" provider option, point it at https://api.libertai.io/v1 with your API key and a model id from the table below. That is the whole integration.
Get an API key
Sign in to the Developer console, create a key, and top it up. If you use the LibertAI CLI, libertai login walks you through browser sign-in and libertai keys create mints keys from the terminal. You can also skip keys entirely and pay per request in USDC on Base with x402 payments.
Check that a key works without spending tokens:
curl -i https://api.libertai.io/libertai/auth/check \
-H "Authorization: Bearer YOUR_API_KEY"200 OK means you are good to go; 401 means the key is missing or invalid.
Models
Recommended default for coding agents: qwen3.5-122b-a10b (262k context, vision, tools). Budget option: qwen3.6-35b-a3b. For confidential inference in a Trusted Execution Environment: hermes-3-8b-tee (note the 16k context).
| Model | Input / output (per M tokens) | Context | Capabilities |
|---|---|---|---|
hermes-3-8b-tee | $0.15 / $0.60 | 16k | tools, TEE 🔒 |
gemma-4-31b-it | $0.15 / $0.40 | 262k | vision, tools |
qwen3.6-27b | $0.15 / $0.50 | 262k | vision, tools |
qwen3.6-35b-a3b | $0.15 / $0.50 | 262k | vision, tools |
qwen3.5-122b-a10b | $0.25 / $1.75 | 262k | vision, tools |
deepseek-v4-flash | $0.25 / $1.75 | 200k | tools |
Thinking variants exist with a -thinking suffix (for example qwen3.6-27b-thinking). The live, authoritative list with current pricing is on the Text API page, or via GET /v1/models.
See also
- Quickstart — first API call in under five minutes
- Text usage — streaming, function calling, vision, direct CRN access
- Trust model & TEE — what the 🔒 flag guarantees
- x402 payments — pay per request without an API key