Tencent unveils WeLM-Image: 31.7B MoE image model tops open-source benchmarks, 5s per image on one H20

Crazy-Repeat-2006 · reddit · 2026-11-08

Tencent/WeChat released the WeLM-Image paper, a unified image foundation model combining multimodal understanding with text-to-image, instruction-guided editing, and multi-reference synthesis, designed for single-GPU deployment in WeChat's consumer apps.

Architecture

Mixture-of-Transformers backbone with shared attention and separate understanding/generation parameters; per-path MoE modules give 31.7B stored parameters with 5.4B activated per target latent token
Text, reference images, and target latents processed as one interleaved sequence; progressive training up to native 1.5K resolution
A knowledge-grounded visual vocabulary plus large-scale visual clustering balances semantics and editing tasks

Training and results

Rubric/checklist-based rewards optimize generation and editing separately, consolidated via noise-aware on-policy distillation
The 1536-resolution model ranks first among evaluated open-source models on PRISM-Bench, Qwen-Image-Bench, and UnicBench
MeanFlow few-step distillation yields WeLM-Image-Turbo: 1024×1024 in 4.98 s on a single H20 with 8 guidance-free steps

Original post →

Tencent unveils WeLM-Image: 31.7B MoE image model tops open-source benchmarks, 5s per image on one H20

More from Multimodal