LongCat team (Xu Huang et al.) · Compressed AR Image
G-STAR: Group-Scaled Token Capacity for Deeply Compressed Autoregressive Generation
G-STAR is a spatial-group co-scaling framework for discrete autoregressive image generation. It enables 32× and effective 64× spatial compression while keeping strong reconstruction, understanding and generation performance, and adds a semi-autoregressive decoder to manage the latency cost of larger groups. The authors report competitive results on GenEval, DPG-Bench, PRISM, LongText-Bench and Qwen-Image-Bench, with a 20× speedup over prior discrete autoregressive models and about 2.5 seconds per 1024×1024 image on a single NVIDIA H200.
Open the source ↗Similar Vision & Image releases
- Seedream 5.5
ByteDance Seed · Image generation and editing · Week 47, 2026
- MiniMax-H3-Image
MiniMax (MiniMaxAI) · Open 16MP Image Model · Week 47, 2026
- Pyramid-JiT (P-JiT) 2B
Linum · Pixel-space T2I · Week 46, 2026
- P-H3-Image
Pruna AI · Open image model · Week 46, 2026
- NitroFusion2
Dar-Yen Chen and Yi-Zhe Song · One-step diffusion distillation · Week 46, 2026
- Noct Q: Qwen Image 2.1 Uncensored Realism (V4)
Noctaluna · Uncensored image model · Week 46, 2026
Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.