Developer Docs
Developer documentation
Connect to the AI gateway with OpenAI-compatible calls, authenticated requests, failover, semantic cache, and PII masking.
Quickstart
coolbestllm is compatible with OpenAI-style clients. Change the Base URL to enable routing, failover, cache, and masking.
export OPENAI_BASE_URL="https://api.coolbestllm.com/v1" export KEY="YOUR_API_KEY"
Authentication
Send Authorization: Bearer <KEY> with every request. Keys can be scoped by model allowlist and token budget.
Unified endpoints
| Endpoint | Description |
|---|---|
/v1/chat/completions | Chat completions compatible with OpenAI clients |
/v1/embeddings | Vector embeddings |
/v1/models | List available models |
Failover
Configure ordered backup routes for each model. If the primary channel fails, the gateway uses the next approved route.
Semantic cache
Enable cache by request header to reuse approved responses for semantically similar prompts.
PII masking
The gateway can detect names, payment fields, IDs, and other sensitive values before forwarding prompts upstream.