AI
Claude's own maker — strong reasoning and long context, built with more thought given to safety than most.
Enterprise-leaning APIs built around retrieval and search, not demo flash — a solid pick for RAG-heavy products.
Google's answer, with real native multimodal input and a free tier generous enough to actually prototype on.
Not a model maker — custom chips that run open models like Llama at speeds that make streaming feel instant.
European and more open about its weights than the US giants, usually cheaper too if that matters to you.
The default LLM API for a reason — GPT models, huge ecosystem, and every tutorial assumes you're using it.
One API key, hundreds of models from every provider — the sane way to avoid betting your app on a single vendor.
Cheap, fast hosting and fine-tuning for open-weight models, for teams who'd rather not manage their own GPUs.