Speridlabs Releases Iris-3B: Pixel-Space Generative Model That Can Replace DINOv2
Apprehensive_Sky892 · reddit · 2026-10-09
Speridlabs Research released Iris-3B, a generative model that works directly in pixel space as a general vision learner.
- It bypasses the lossy compression and texture-biased latents of a VAE, helping on dense tasks where detail matters.
- The team shows image generators can replace discriminative vision models like DINOv2 by putting the generative prior to work on dense prediction.
- Pixel-space scaling recipes are published; a 12GB fp32 checkpoint is available on Hugging Face.
Related event: Sperid Labs Open-Sources Pixel-Space Text-to-Image Model Iris-3B(4 posts)→
More from Multimodal
- GPT-6 Luna Shows Surprising Skill at Spotting AI-Generated Images — Angaisb_ · 2026-10-09
- KAIST's LongTake: long-horizon teacher forcing keeps 30-60s video generation dynamic — kaist-ai · 2026-10-09
- AI image of the day: GPT-6 Image 2.5 recreates 1940s Chinese village newsreel — DeryaTR_ · 2026-10-09
- Video character swap on 8GB VRAM: 640x480 render in 10:33 on RTX 4070 — big-boss_97 · 2026-10-09
- Google opens SynthID Detector globally; Vidu Q4 Preview prices video gen from $0.013/sec — 创业邦 · 2026-10-09
- Vigglorious Studio: chunked reference frames and guide keyframes fix character drift in long AI videos — Tablaski · 2026-10-09