LocalLLaMA
AI Benchmark Race Spirals Out of Control — Developers Skeptical of Scores
LocalLLaMA post sparks debate: hundreds of new AI benchmarks launched yearly, but model scores diverge from real capability, looking like marketing to
Back to home
model selection
2 articles tagged with this topic
Qwen3
Qwen 3 还是 Gemma 4？本地 部署玩家正在用实测替 代官方跑分——小模型选型 进入「场景优先」时代
A Reddit thread comparing Qwen 3 35B and Gemma 4 26B reveals a shift: users now trust personal testing over official benchmarks.