Model
DiffusionGemma 26B MoE
A diffusion model based on Gemma that adopts an MoE architecture, has 26B parameters, and features a high degree of parallelism.
Timeline (last 90 days)
Oct 17
On 8×B200, DiffusionGemma 26B MoE + ORL reaches 7,708 tokens/s at NVFP4 precision (single-request decoding).
Unverified50%