Xiaomi MiMo Releases DFlash Acceleration Scheme
karminski3 · x · 2026-07-13
Xiaomi released **MiMo-V2.5-DFlash** on Hugging Face; the code is already out even though the README hasn't been updated yet. The focus isn't on "how strong the new model is," but rather its **speculative decoding implementation**: the draft model uses **block diffusion** to guess an entire block of tokens in a single forward pass, which are then verified by the large model all at once. The author breaks down the code, noting that this is a completely different route from traditional autoregressive drafting like EAGLE. Implementation details mentioned: - In coding contexts, acceptance length can reach 6+, resulting in significant speedups - Draft weights are extracted separately, taking up about 2.94G locally - It's not a standalone small model; lacking embed / lm_head, it still requires KV injection from the main model's hidden state - `target_layer_ids` are aggressively set to `[0, 11, 23, 35, 47]` - `block_size=8` is more conservative than in papers and Qwen implementations, potentially offering more stability but slightly lower parallelism Overall, this is a technical teardown focused on an **inference acceleration plugin/deployment scheme** rather than just a new model announcement.
Related event: Xiaomi Releases MiMo-V2.5-DFlash Weights and Code(2 posts)→
More from Infra
- UK AI datacentres face backlash over heat, noise and land use — nordicinst · 2026-07-21
- Fluidstack raises $830M at $7.5B valuation as Anthropic backs a $50B compute buildout — rohanpaul_ai · 2026-07-21
- Early Krea2 Gradio WebUI targets 6GB low-VRAM local runs — Fluid_Kaleidoscope17 · 2026-07-21
- Z.AI starts running a 1GW AI data center built entirely on domestic chips — Polymarket · 2026-07-21
- Local models feel far more capable once paired with the right harness — Soft-Barracuda8655 · 2026-07-21
- Voice-agent teams should use platforms first, then own STT events when failures get weird — FollowingSuitable941 · 2026-07-21