Documentation
NexoRouter documentation for OpenAI-compatible API setup, dashboard workflows, billing, errors, and integrations.
NexoRouter Docs
NexoRouter is an OpenAI-compatible AI API gateway for developers and small teams. Use one NexoRouter API key to call publicly available models, manage prepaid balance, test prompts in Playground, and inspect usage logs.


Start here
2-minute quickstart
Create an API key, set https://api.nexorouter.com/v1 as the base URL, choose a model ID from Models, and send a first chat/completions request.
Read next: Quickstart
Dashboard workflows
Learn how API key budgets, expiry, model scope, key reveal, revoke, Playground, Billing, and Usage Logs work together.
Read next: Dashboard workflows
Tool integrations
Use NexoRouter in tools that support OpenAI-compatible chat completions. Treat candidate setups as trial-only until their page says they are verified.
Read next: Tool integrations
API reference
Use the documented stable endpoints, request body rules, error shape, rate limits, and endpoint boundaries.
Read next: API Reference
API capabilities
Understand which API shapes are stable today and which require verification before production use.
Read next: API Capabilities
Status and incidents
Check Status, Usage Logs, and request IDs before deciding whether a failure is local, account-related, gateway-related, or upstream.
Read next: Status and incidents
Copy these first
Base URL: https://api.nexorouter.com/v1
Header: Authorization: Bearer YOUR_NEXOROUTER_API_KEY
Starter model: deepseek-v4-flash
Public API surface
Only the endpoints listed in the API reference are part of the stable public support commitment. Other /v1 paths may pass through the gateway, but they are not documented product guarantees and can fail or change until they are verified here.
| Capability | Public status |
|---|---|
GET /v1/models | Supported. Returns publicly available model IDs. |
POST /v1/chat/completions | Supported for OpenAI-compatible chat requests. |
| Streaming | Not documented as a verified stable public feature yet. |
| Embeddings | Not documented as a verified stable public endpoint yet. |
| OpenAI Responses API | Not part of the documented stable public API yet. |
| Anthropic Messages API | Not part of the documented stable public API yet. |
| Gemini native API | Not part of the documented stable public API yet. |
Common tasks
- Create an API key.
- Create a key with the right budget, expiry, or model scope.
- Move an OpenAI SDK app to NexoRouter.
- Pick a model ID from Models.
- Choose a first model and check model availability, versions, aliases, and request cost.
- Control output length with
max_tokens. - Test a prompt in Playground.
- Check tokens, cost, latency, and request IDs in Usage Logs.
- Fix
invalid_api_key,model_not_found,insufficient_quota, rate limits, timeouts, and account access issues.
Integration status
| Tool | Status | Why |
|---|---|---|
| OpenAI SDK | Supported public API | Standard OpenAI-compatible chat completions. |
| OpenCode | Candidate setup, not yet verified by NexoRouter | OpenCode supports custom OpenAI-compatible providers for /v1/chat/completions. |
| Cline | Candidate setup, not yet verified by NexoRouter | Cline supports OpenAI-compatible providers. |
| Roo Code | Candidate setup, not yet verified by NexoRouter | Roo Code supports OpenAI-compatible providers but requires model tool-calling compatibility for agent tasks. |
| Aider | Candidate setup, not yet verified by NexoRouter | Aider can use OpenAI-compatible endpoints through OpenAI-compatible configuration. |
| Continue | Candidate setup, not yet verified by NexoRouter | Continue supports OpenAI-compatible chat model configuration. |
| Cursor | Candidate setup, not yet verified by NexoRouter | Cursor can work only when its current settings expose a custom OpenAI-compatible endpoint. |
| Open WebUI | Candidate setup, not yet verified by NexoRouter | Open WebUI supports OpenAI-compatible connections and model allowlists. |
| LangChain | Candidate setup, not yet verified by NexoRouter | ChatOpenAI can use a custom base_url. |
| Vercel AI SDK | Candidate setup, not yet verified by NexoRouter | @ai-sdk/openai-compatible accepts baseURL and apiKey. |
| Dify | Candidate chat setup, not yet verified by NexoRouter | Dify can use OpenAI-compatible model providers; embeddings and rerank need separate checks. |
| LlamaIndex | Candidate chat setup, not yet verified by NexoRouter | Chat calls can use an OpenAI-compatible base URL; embeddings remain separate. |
| Translation tools | Candidate setup pattern, not yet verified for specific clients | Only works when the client exposes an OpenAI-compatible custom base URL. |
| Community integrations | Not verified yet | Use the compatibility checklist before trying plugins or workflow tools. |
| Codex CLI | Not supported by current public API | Current Codex provider configuration expects Responses API behavior; NexoRouter's stable public API is Chat Completions. |
| Claude Code | Not supported by current public API | Claude Code routes through an Anthropic-compatible LLM gateway, not an OpenAI-compatible /chat/completions endpoint. |
| Gemini CLI | Not supported by current public API | Gemini CLI expects Gemini-native behavior unless a current version adds a compatible mode. |
When something fails
Start with the error code and the Dashboard:
invalid_api_key: check the header, key value, expiry, and enabled state.model_not_found: copy the exact model ID from Models and check the key's model scope.- Unexpected model cost: check model ID, prompt tokens, completion tokens, retries, and agent loops in Usage Logs.
insufficient_quota: check Billing and the key budget.rate_limit_exceededortoken_rate_limit_exceeded: wait for the retry window or reduce request volume.request_too_large: reduce the input size; retrying the same payload will not succeed.upstream_request_timeout: increase your client timeout or try another model.