# ZipDepth Distills a Vision Foundation Model Into a 6-Million-Parameter Depth Network

The compact model converts into plain convolutions with no custom operators, keeping its zero-shot accuracy while running in real time on GPUs, CPUs, and embedded hardware.

- Published: 2026-07-11T06:01:51.690Z
- Canonical: https://polylog.news/ai/2026-07-11/zipdepth-distills-a-vision-foundation-model-into-a-6-million
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [arXiv](https://arxiv.org/html/2607.08771v1)

Researchers at the University of Bologna released ZipDepth, a network that estimates depth from a single camera image and transfers the broad knowledge of a large foundation model, Depth Anything v2-Large, into a smaller model. The result i…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-07-11/zipdepth-distills-a-vision-foundation-model-into-a-6-million (subscription information: https://polylog.news/pricing).