LLM Providers & APIs Tools

This category is the supply chain of AI: the companies that train frontier models and the platforms that serve them. It's here for builders — if you're an end user, you want a chatbot; if you're shipping an AI feature, this is where it comes from.

Three groups matter. First, the frontier labs selling API access to their own models: OpenAI API, Anthropic API, Google AI Studio, Mistral AI, Cohere, and AI21 Labs. You get the newest models first, direct from the source. Second, aggregators and marketplaces: Amazon Bedrock serves multiple model families inside AWS (useful if your infrastructure and compliance already live there), and OpenRouter gives you one API key for dozens of models — the fastest way to compare and switch without re-integrating. Third, inference specialists: Groq, Together AI, Fireworks AI, and DeepInfra host open-weight models — including Meta Llama, the open-weights anchor — optimized for speed and cost.

The differentiators are unglamorous and decisive: price per million tokens, latency and throughput, context window, rate limits, and fine-tuning options. Pricing is pay-as-you-go almost everywhere, which makes unit economics easy to test and easy to get wrong at scale. The practical advice: prototype through an aggregator like OpenRouter, then move production traffic to whichever provider wins on your actual workload's cost and latency — not on benchmark charts.