Integrations

Agents & OpenAI-compatible clients

Point any OpenAI-compatible client or agent framework at Roteo.

Roteo's /v1/chat/completions and /v1/responses endpoints work with any OpenAI-compatible client: chat apps, coding agents and agent frameworks.

Universal settings

Setting	Value
Base URL	`https://api.roteo.ai/v1`
API key	`sk_your_key`
Chat path	`/chat/completions`
Model	an open-weight or `gpt-*` id, e.g. `glm-5.3` or `gpt-5.6-sol`

For agents, use an open-weight or GPT model. Agent frameworks are refused on Claude models, with a message telling you to switch; Claude works in Claude Code and in direct API calls.

Model support for agents

Models	opencode	Other agent frameworks (Hermes, OpenClaw, Cursor)
Open-weight: `glm-5.3`, `glm-5.3-flash`, `deepseek-v4-flash-0731`, `kimi-k3`, `qwen3.8-max`	Recommended, verified with real tool calls	Supported, not yet verified per framework
GPT: `gpt-*`	Supported, not yet verified	Supported, not yet verified per framework
Claude: `claude-*`	Not available	Not available; use Claude Code

Guides per tool

Chat apps

Any app with a "Custom Provider" or "OpenAI-compatible" option works: Chatbox, LobeChat, Cherry Studio, NextChat, Open WebUI and more. Use the universal settings above; most apps load the model list from GET /v1/models on their own.

Anything else

If it speaks the OpenAI Chat Completions or Responses protocol, it works: LangChain, LiteLLM, Continue, Aider and the OpenAI SDKs. See the OpenAI SDK guide. An agent that can read can also onboard itself from /v1/llms.txt.