Community Drafter Lifts Qwen3.8-27B NVFP4 Decode to 48 tok/s on DGX Spark

A community-tuned drafter for Qwen3.8-27B's NVFP4 build raises DGX Spark decode speed from 12.5 to 48.2 tokens per second, a near-4x gain, while remaining bit-exact lossless.

Read the full story at AlphaSignal ↗

Timeline · 1 report

2026-11-01 22:00 · AlphaSignal
AutoTrust's Qwen3.8 Drafter Hits 48 tok/s, Nearly 4x Faster on DGX Spark

More stories

AutoTrust Releases Open JEV-9B Model for Fast AI Decisions — AlphaSignal
OpenAI plans post-midterms PR blitz, rebranding data centers as 'AI compute campuses' — Tom's Guide AI
Tech stocks slide as AI spending concerns mount — Financial Times AI
Saudi Aramco Acquires 49% Stake in AI Startup Humain for ~$2.6B — Techmeme
Stanford HAI Fellow Raises Concerns Over AI Systems Inferring User Data Without Disclosure — Stanford HAI
DeepSeek open-sources dsh-superagent, a proactive coding agent prototype — Bloomberg AI
Opinion | China and US need to join hands against another powerful adversary: AI — South China Morning Post Tech
Anthropic updates Claude to stop it from recreating existing video games — Engadget

Get the daily brief of stories like this at 6:30 every morning →