LLM APIs and model providers · ranked 02
OpenAI API
The default model API. GPT models, tools, and a platform console.
Best for The model API most libraries already speak
The OpenAI API is the platform for OpenAI's models, including GPT models, tool use, and related APIs. It is the default when a tutorial, an SDK, or a teammate already assumes an OpenAI key. Billing is usage-based. Rates: check site.
Best use cases
- Adding a model call to an app that expects an OpenAI-style API
- Tool calling and structured outputs
- A key you can also use from Codex and many frameworks
Pros
- + The widest support in frameworks and examples
- + Usage-based, so a small app can start small
- + One platform covers the models and the coding agent
Cons
- − The bill tracks usage, and usage on agents adds up
- − Rates, model names, and limits: check site
Pricing
Usage-based. Token rates and rate limits: check site.
Alternatives
LLM APIs and model providers
8.3
Anthropic API
Claude models via API. The pair to Claude Code.
Best for Calling Claude from your own product
LLM APIs and model providers
7.6
Gemini API
Google's Gemini models, with a free tier that moves.
Best for A capable model API with a free starting quota
LLM APIs and model providers
7.9
OpenRouter
One API key for many models, including labs and open weights.
Best for Switching models without switching SDKs