Client setup¶
All clients use the same Nexus key and key-specific model catalog. The base URL depends on the protocol:
| Workflow | Protocol | Base URL | First endpoint |
|---|---|---|---|
| Codex and response clients | OpenAI Responses | https://api.nexus-hub.ru/v1 |
POST /v1/responses |
| Claude Code | Anthropic Messages | https://api.nexus-hub.ru |
POST /v1/messages |
| Generic OpenAI clients | OpenAI Chat | https://api.nexus-hub.ru/v1 |
POST /v1/chat/completions |
| Gemini clients | Gemini-compatible | https://api.nexus-hub.ru/v1beta |
POST /v1beta/models/{model}:generateContent |
General setup¶
- Create a Nexus API key in the Telegram bot.
- Select a separate profile or custom endpoint in the client.
- Enter the protocol and base URL from the table above.
- Store the key in the client's protected field or an environment variable.
- Select a model returned by
GET /v1/models.
Choose a protocol¶
Use it for Codex and response clients. Set the base URL to
https://api.nexus-hub.ru/v1 and begin with POST /v1/responses.
Use it for Claude-compatible clients. Keep the base URL at
https://api.nexus-hub.ru and send POST /v1/messages with
anthropic-version: 2023-06-01.
Use https://api.nexus-hub.ru/v1beta only for clients that document the
Gemini-compatible format. The model belongs in the request path.
Before troubleshooting
Recheck the endpoint, model visibility for this key and exact client
setting. Send support only the route, error code, approximate time and
x-request-id; never include the key, a full prompt or request headers.
Start with the client parameters. Dedicated guides cover Codex CLI, ChatGPT / Codex Desktop, Claude Code, Claude Desktop, VS Code, Zed, T3 Code, CC Switch, Direct API, custom integration, and other clients.