Tencent's DSpine draft model lifts Qwen3.8-27B speculative decoding to 3x, 805 tps
karminski3 · x · 2026-11-10
Tencent Youtu released Qwen3.8-27B-DSpine-b16 on Hugging Face, boosting speculative decoding from 1.6x (standard MTP) to 3x, with official tests hitting 805 tps at 8-way concurrency; FlashAttention 3 suggests Hopper GPUs.
Related event: Tencent's DSpine Speeds Up Qwen3.8-27B Inference 3x(3 posts)→
More from Infra
- "Biblically accurate" low-loss waveguide crossing looks majestic, performs terribly — jwt0625 · 2026-11-10
- RBA gives LLaVA-OneVision-2 a 3x lossless training speed-up, merged upstream — liuziwei7 · 2026-11-10
- Comet 60B (A12B) gets GGUF quants with MTP head for speculative decoding in llama.cpp — Felladrin · 2026-11-10
- Nvidia lobbied hard against rules on remote access to restricted chips, officials say — rohanpaul_ai · 2026-11-10
- AI daily: Fable 5.2 tops evals, orchestrates 1000 agents for 8 days to build a prototype OS — jiqizhixin · 2026-11-10
- Engineer calls megakernel hype a psyop: CUDA graphs already express a strict superset — mike64_t · 2026-11-10