Skip to main content

​
Welcome to Openference

Openference is a curated AI model proxy. It provides a single, stable, OpenAI-compatible API endpoint that intelligently routes your requests to the best available upstream providers. One base URL. One API key. Many models.

​
Why Openference?

  • Drop-in compatible: Works with Cursor, Claude Code, Codex, Continue, Cline, OpenAI SDKs, LiteLLM, LangChain, and any OpenAI-compatible client.
  • Curated catalog: Hand-picked high-quality models across providers. We handle deprecations so you don’t have to.
  • Automatic failover & key rotation: If one upstream provider is down or rate-limited, we route to healthy alternatives.
  • Transparent pricing: Per-million-token input/output prices shown for every model. No surprise bills.
  • Usage separation: Plan quotas vs pay-as-you-go credits are tracked distinctly.
  • Format translation: Send OpenAI format to Anthropic models (and vice-versa). We convert for you.
  • Fast & global: Runs on Cloudflare’s edge network for low latency worldwide.

​
Supported API surfaces

EndpointFormatNotes
POST /v1/chat/completionsOpenAIPrimary chat interface, streaming, tools
POST /v1/messagesAnthropicClaude-style messages
POST /v1/responsesOpenAIResponses API (Codex etc.)
POST /v1/embeddingsOpenAIText embeddings
GET /v1/modelsOpenAILive model list (filtered by your key’s restrictions)
All endpoints are available at: 
https://api.openference.com/v1

​
Get started in 60 seconds

  1. Register and verify your email.
  2. Log in and create an API key (starts with sk-token-).
  3. Send your first request:
curl https://api.openference.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-token-YOUR_KEY" \
  -d '{
    "model": "your-model-name",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'
See the Quickstart for full walkthroughs with popular clients.

​
Next steps

Questions? Check the FAQ or Troubleshooting.