API Guides
These guides answer “how do I use this API?” Choose a protocol or capability, then follow the request flow and examples. Every request requires an API key and a compatible model ID.
If you need exact fields, parameter types, response schemas, or status codes, use the API Reference.
Text and agents
- Chat Completions — OpenAI-compatible message-based text and multimodal requests
- Responses API — response items, reasoning, tools, and continuation workflows
- Streaming — receive incremental output from Chat Completions or Responses
Retrieval
- Embeddings — convert text into vectors
- Rerank — score and reorder documents against a query
Media
- Images — generate and edit images
- Audio — transcribe, translate, and synthesize speech
- Video — create and poll asynchronous video tasks
Provider-compatible protocols
- Claude Messages — Anthropic-compatible message requests
- Gemini-compatible API — Gemini model listing and
generateContent
Before sending a request
- Create an API key.
- Copy an exact model ID from Models.
- Verify that the model supports the endpoint and capabilities you need.
- Review the live Pricing page.
The API Reference is the machine-generated contract for the public endpoints; it complements these task-oriented guides.
