Google · 26B (4B active) · 專家混合（MoE）
Discrete diffusion MoE — 1100+ tok/s on H100, multimodal (text/image/video) 看看你的 GPU 或 Mac 跑不跑得動 DiffusionGemma 26B-A4B IT——最低 14.5 GB，建議 24.2 GB。
2026-06256K context
專家混合（MoE）
專家總數： 128
啟用專家： 8
啟用參數： 3.8B
| 量化 | 位元 | VRAM | 品質 | 狀態 |
|---|---|---|---|---|
| Q2_K | 2 | 8.8 GB | low | — |
| Q3_K_M | 3 | 12.2 GB | moderate | — |
| Q4_K_M | 4 | 13.8 GB | good | — |
| Q5_K_M | 5 | 17.1 GB | good | — |
| Q6_K | 6 | 20.5 GB | excellent | — |
| Q8_0 | 8 | 27.1 GB | excellent | — |
| F16 | 16 | 53.8 GB | lossless | — |
Can I run DiffusionGemma 26B-A4B IT locally?
- Can I run DiffusionGemma 26B-A4B IT locally?
- DiffusionGemma 26B-A4B IT needs about 14.5 GB of memory at a minimum and 24.2 GB recommended. Open this page to grade it against your GPU or Mac, then run it with runai, Ollama or LM Studio.
- How much VRAM does DiffusionGemma 26B-A4B IT need?
- At Q4_K_M, DiffusionGemma 26B-A4B IT uses about 13.8 GB of VRAM. Higher quants need more memory; lower quants fit tighter cards with a quality tradeoff.