Qwen-4-max-preview Scores Below Qwen-3.8-flash in Early User Benchmarks
teortaxesTex · x · 2026-11-11
A developer's GameFeel RE benchmark runs show Qwen-4-max-preview scoring worse than qwen-3.8-flash, qwen-3.8-max, qwen-4-27b-early and qwen-4-flash-early — while being the second-cheapest model tested (preview discount?). Another user adds that the checkpoint "gives straight-up 7B model vibes in pretty much every task," hoping it's just a rushed misconfigured release. The community is puzzled by the flagship preview's quality.
More from Models
- fable 5.2 discovers ElevenLabs MCP on its own, turns two Midjourney images into an audio story — ciguleva · 2026-11-11
- Same photo test: fable 5.2 nails image-to-illustration where opus 5.5 fell short — yuwen_lu_ · 2026-11-11
- User hits Qwen4-Max usage cap after ~1 hour, suspects lowered limits vs Qwen3-Max — cedric_chee · 2026-11-11
- Unverified speculation: Opus 5.5 allegedly a 'diluted' Fable 5.5; stronger model too unusable to ship — haider1 · 2026-11-11
- Fable 5.2 builds a 3D game in 76 min for ~$42 in tokens, roughly matching Opus 5.5 — FinanceYF5 · 2026-11-11
- Security researcher: Abliterated models drastically improve offensive-security reasoning — matthew_d_green · 2026-11-11