Provider pricing

Baseten AI API pricing

Baseten is a production inference platform purpose-built for serving open-weight and custom models at scale. Model APIs cover Llama 4, Qwen3, DeepSeek V4, and Whisper at competitive per-token rates, while Truss-packaged custom models run on autoscaling H100/H200 capacity with cold-start optimization, multi-region failover, and observability built in.

Founded2019
HeadquartersSan Francisco, CA
Tracked modelsDirectory

Best for

Teams shipping custom models or fine-tunes to production with strict latency and uptime requirements.

Strengths

  • Production-grade autoscaling with sub-second cold starts
  • Truss: deploy any custom model from a Python class
  • Multi-region, multi-cloud failover for high availability
  • Per-token Model APIs and per-second dedicated deployments on the same platform