Deploy the best open-weight AI models on dedicated GPUs — OpenAI-compatible APIs, billed by the second, terminate anytime.
Any HuggingFace model works — paste its ID in the deploy form and we'll serve it with SGLang or vLLM.
