Qwen-4-27B early checkpoint tops GameFeel Bench, best open-source result by far
teortaxesTex · x · 2026-11-08
Developer follistatindev was granted access by Alibaba and added an early checkpoint dubbed 'QoDuck' (qwen-4-27b) to GameFeel Bench (60k game-balancing and feature-understanding tasks). It is beaten only by Opus 5.5-Max and Xhigh — far ahead of every open-weight model. The size-vs-performance chart shows the 35GB (quantized) 27B scoring 47, well above Deepseek V4.1-Flash-1017 (42.8 at 800GB) and Kimi K3.1-Pro (30 at 3000GB). teortaxesTex notes that if the benchmark correlates even slightly with overall strength, Alibaba's next release could be brutal.
More from Models
- Ampersand, a Model Built for Group Chats, Free on Vercel AI Gateway Through Nov 13 — gharik · 2026-11-08
- Observer Notes Inkling-2-Small's CoT Sounds a Lot Like Newer GPT's — Sauers_ · 2026-11-08
- LiquidAI ships official NVFP4 quantized weights for LFM3-1.7B and VL-2B — JosephJacks_ · 2026-11-08
- 'Rank bottleneck' means two very different things in linear algebra vs systems optimization — kalomaze · 2026-11-08
- Double-exponential fit suggests model time horizons could hit ~80 work-years by end of 2027 — scaling01 · 2026-11-08
- Speculation: Anthropic's recent releases are all distills, Opus 5.5 may be Model 2 Early — haider1 · 2026-11-08