ZenLLM Docs

ZenLLM stays close to OpenAI-compatible request shapes, so the common migration path is simple: swap the base URL, use your ZenLLM key, and choose the model you need.

All requests require an API key. Pass it via the appropriate header depending on the API format you're using.

OpenAI format

Authorization: Bearer sk-zenith-...

Anthropic format

x-api-key: sk-zenith-...

Use the appropriate base URL for your client's API format.

OpenAI

https://api.zenllm.org/v1

Anthropic

https://api.zenllm.org/v1

Responses

https://api.zenllm.org/v1

Core endpoints

OpenAI-compatible chat completions. Works with Cursor, Cline, Aider, OpenCode, and any OpenAI SDK.

Non-streaming

curl https://api.zenllm.org/v1/chat/completions \
  -H "Authorization: Bearer sk-zenith-..." \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-5.5","messages":[{"role":"user","content":"hello"}]}'

Streaming

curl -N https://api.zenllm.org/v1/chat/completions \
  -H "Authorization: Bearer sk-zenith-..." \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-5.5","messages":[{"role":"user","content":"hello"}],"stream":true}'

Streaming emits SSE data: lines with chat.completion.chunk objects and ends with data: [DONE].

Newer OpenAI Responses API format. Used by Codex CLI, OpenAI Agents SDK, and OpenAI SDK v5+.

curl https://api.zenllm.org/v1/responses \
  -H "Authorization: Bearer sk-zenith-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5.5",
    "input": "hello",
    "instructions": "You are a helpful assistant."
  }'

Uses named SSE events (event: response.created, response.output_text.delta, etc.). No [DONE] sentinel.

Anthropic Messages API format. Used by Claude Code and the Anthropic SDK.

curl https://api.zenllm.org/v1/messages \
  -H "x-api-key: sk-zenith-..." \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5.5",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "hello"}]
  }'

Uses Anthropic SSE events (event: message_start, content_block_delta, etc.). Requires anthropic-version: 2023-06-01 header.

List available models. Every client calls this for model discovery.

curl https://api.zenllm.org/v1/models \
  -H "Authorization: Bearer sk-zenith-..."

Browse all available models on the Models page.

Generate vector embeddings for text, for semantic search and RAG.

curl https://api.zenllm.org/v1/embeddings \
  -H "Authorization: Bearer sk-zenith-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/text-embedding-3-large",
    "input": "Your text to embed"
  }'

Available models

Rerank documents by relevance to a query.

curl https://api.zenllm.org/v1/rerank \
  -H "Authorization: Bearer sk-zenith-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3-reranker-8b",
    "query": "What is the capital of France?",
    "documents": ["Paris is the capital of France.", "London is the capital of the UK."]
  }'

Available models

Stuck on an integration, hit an unexpected error, or have a billing question? The fastest way to reach the team and community is Discord. We're active and usually reply within a few hours.