Apple Depth Pro 2: sharp, temporally consistent metric depth for images and video
Apple released Depth Pro 2, a family of four open-weight models (24M-872M) that estimate metric depth in meters from a single image, video clip or live camera feed, with a separate focal-length estimator and a research-only license.
Why it matters
One model handles stills and video at full resolution with both sharp boundaries and frame-to-frame consistency, the trade-off previous monocular depth methods forced, and the smallest model runs at 53 fps at 1080p.
What you could build with it
A small team could build a phone app that turns a short video walkthrough of a room into a measured 3D floor plan, using Depth Pro 2's metric depth to size furniture placement without a lidar sensor.
Does it hold up?
Too early to judge: released Nov 19-20 with code and weights public, but the research-only license limits commercial use and no third-party production builds have surfaced yet.
Built with Apple Depth Pro 2
- Depth Pro 2: Sharp and Temporally Consistent Metric Monocular Depthalphaxiv · Paper breakdown with benchmark numbers: H+ scores 41.4% on delta-1 on Sintel video with predicted intrinsics versus 37.0% for UniDepthV2-L, and the smallest model runs at 53 fps at 1080p.
- Apple Depth Pro 2: open depth model for photo and videoinite.ai · News coverage of the release: four model sizes, a focal-length estimator, video support and the research-only license.
Learn more
First spotted on article: source.
More AI models
MiniMind: train your own 64M LLM from scratch in about 2 hoursA complete, from-scratch LLM training pipeline at 64M parameters - pretraining, SFT, LoRA, DPO, PPO/GRPO, tool use and…models · JEV 0.82VoiceStudio: the open-source, fully-local ElevenLabs alternativeA desktop app bundling 20 local speech-generation engines and 10 transcription options behind one UI — voice cloning…models · JEV 0.76Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73tiny-voca: a fully open 4.5M-parameter TTS trained for under $10 of computeA fully open 4.5M-parameter English TTS — code, weights and dataset under Apache 2.0 — trained from scratch in 7.7…models · JEV 0.67Qwen4-Max: Alibaba's new flagship model goes liveAlibaba's 2.4T-parameter flagship is now generally available via API, with a 1M-token context, native vision, and…models · JEV 0.67
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.