Provider pricing

Moondream AI API pricing

Moondream is a 2B-parameter open-weight vision-language model optimized for edge and high-throughput cloud inference. It supports VQA, captioning, object detection via pointing, and OCR at a fraction of the cost of GPT-5 Vision or Claude Vision. The hosted Cloud API is OpenAI-compatible with per-image pricing and a free local SDK for on-device use.

Founded2024
HeadquartersSeattle, WA
Tracked modelsDirectory

Best for

Edge devices, embedded vision pipelines, and cost-sensitive batch image processing.

Strengths

  • 2B params — runs on a single consumer GPU or even CPU
  • Pointing and bounding-box outputs for grounded vision tasks
  • Cloud API priced per image, well below GPT-5 Vision
  • Apache 2.0 weights — fully self-hostable