Editor's pick
Groq
Extremely fast inference for open models.
What it is
It uses its own hardware (LPU) to generate tokens far faster than a traditional GPU. It serves popular open models through an OpenAI-compatible API, which makes migrating existing code trivial.
Best for
Chat interfaces where latency is noticeable.
What the free plan includes
- Request and token limits per minute and per day
- API compatible with the OpenAI SDK
- A catalogue focused on open models
Free plans change often. Confirm the limits on the official page before deciding.
Hugging Face
PickCommunity models, datasets and runnable demos.
Free for public models, datasets and Spaces
Google AI Studio
PickPrototype with the Gemini models from your browser.
Free studio access and a free API tier
NVIDIA NIM
Hosted APIs for AI models optimised by NVIDIA.
Free API credits for prototyping