Documentation

Generate text with LLMs. Streaming and non-streaming.

Generate vector embeddings for search and RAG.

Transcribe audio files to text with Whisper.

Convert text to natural-sounding speech audio.

Generate images from text with DALL-E 3 and GPT Image 1.

Generate videos from text prompts with Sora.

Available models, aliases, and capabilities.

Content safety, PII detection, injection prevention.

Per-tier rate limits and usage quotas.

Use your own API keys for frontier models.

Async bulk inference at 50% off real-time rates.

Deduplicate similar requests for cost savings.

Managed vector search and grounded generation.

Versioned templates with Jinja2 and A/B testing.

A/B testing, budget controls, fallback chains.

WebSocket streaming for audio and text.

HIPAA/SOC 2 exports, retention, legal holds.

System design, inference engines, and scaling.

Deploy ACAI in your own infrastructure.

Full endpoint reference for all APIs.

Environment variables and deployment config.

How to contribute to the open-source project.