Provider pricing

Avian AI API pricing

Avian.io is a high-speed inference provider founded in 2024 and headquartered in San Francisco. Serves Llama and other open models at record tokens-per-second on NVIDIA hardware.

Founded2024
HeadquartersSan Francisco, CA
Tracked modelsDirectory

Best for

Latency-sensitive apps that need fast open-weight inference.

Strengths

  • 351 tokens/sec on Llama 3 70B — fastest NVIDIA-based provider
  • OpenAI-compatible API with no rate limits on paid tier
  • Per-million-token billing with no minimums
  • Listed in the TokenCost AI providers directory