Provider pricing

Novita AI AI API pricing

Novita AI is a serverless inference provider running 200+ open-source models — Llama 4, Qwen3, DeepSeek, FLUX, Stable Diffusion, Wan, and Hunyuan video — at aggressive per-token and per-second rates. The OpenAI-compatible API and broad multimodal catalog make it a popular OpenRouter upstream.

Founded2023
HeadquartersHong Kong
Tracked modelsDirectory

Best for

Multimodal apps and cost-sensitive open-source inference at scale.

Strengths

  • 200+ open-source LLM, image, and video models
  • Aggressive pricing — frequently the cheapest upstream on OpenRouter
  • Per-second GPU billing for image and video generation
  • OpenAI-compatible API with streaming and tool use