Freemium
Hugging Face
The open model hub, and where self-hosted or fine-tuned models come from.
ModelsOne API key for every major model, with per-request routing and cost caps.
OpenRouter puts GPT, Claude, Gemini, Llama and open-weight models behind a single OpenAI-compatible endpoint. Teams building an internal tool can switch models per task, fall back automatically when a provider is down, and see spend per model in one place.
It is infrastructure, not an app - useful only if you are building something. A small markup applies on top of provider pricing.