AI

Claude models via a straightforward Messages API, with prompt caching and batch processing that meaningfully cut cost on repeat or high-volume calls.

Useful once your AI feature needs more than a single prompt, like RAG with multiple retrievers or multi-step agent logic. Overkill for a simple chat wrapper.

The widest model lineup and the most third-party tooling and examples if you want your AI feature to just work with common libraries.

One API key and endpoint that routes to hundreds of models across providers. Handy for A/B testing models or falling back when one is down or rate-limited.

Run open-source and community AI models (image, video, audio) through one API without hosting GPUs yourself.

One TypeScript API for streaming chat/completions across providers, plus ready-made React hooks so you're not wiring SSE by hand.