RT Speridlabs: Meet Iris-3B: a pixel-space T2I model and general vision learner that generates every pixel directly. No VAE. Pre-trained from scratch, 3B, open weights. We also explore its generative prior on detail-critical vision tasks: depth estimation and image restoration. Fully open: weights, code and demo (Apache 2.0)

RT Speridlabs: Meet Iris-3B: a pixel-space T2I model and general vision learner that generates every pixel directly. No VAE. Pre-trained from scratch, 3B, open weights. We also explore its generative prior on detail-critical vision tasks: depth estimation and image restoration. Fully open: weights, code and demo (Apache 2.0) https://speridlabs.com/research/iris

🌐 链接: https://x.com/EMostaque/status/2108297986165866592
🏷️ 分组: X｜正在关注
🕒 时间: 2026-10-09 01:02:38

Speridlabs

Iris-3B · Speridlabs Research

Pixel-space generation & general vision learner — a fully open 3B model with no VAE in the loop, whose generative prior transfers to depth and restoration.