#diffusiongemma

You've probably seen the benchmarks — 4x faster, 6x more mistakes, 273-vote threads on r/LocalLLaMA....

作為三平台評測的最終章（前兩篇為 M2 Max 96GB MLX 與 GH200 vLLM），本篇將完整測試一下 GB10 的吞吐量表現、32K 長 Context 的速度代價、以及在 Podman...

為了找到一些在地端也能讓 Agent 有無限 token 自由的毒駕的方法，原本用手邊的M4 24GB Mac 上嘗試執行 DiffusionGemma 26B，卻悲慘的連 1,000 tokens 的...

DiffusionGemma 26B for Local AI in 2026: 18GB VRAM, 4× Faster Generation, and Which Consumer GPUs Actually Saturate the 1,000 tok/s Ceiling

A build-focused guide to self-hosting Google\'s DiffusionGemma: the exact vLLM serve command, what each diffusion flag does, how to call it like an OpenAI endpoint, and how to tune the speed-vs-qualit

DiffusionGemma hits 1,000 tokens per second by generating text in parallel, but weaker quality keeps it experimental.

Google open-sourced DiffusionGemma on June 10, 2026 — a 26B MoE that writes a 256-token block in parallel instead of one token at a time, hitting 700+ tokens/sec on an RTX 5090 and up to 4x faster tha

AI Dev Weekly is a Thursday series where I cover the week's most important AI developer news, with my...