Companies should build their own benchmarks instead of relying on public leaderboards
simonguozirui · x · 2026-11-10
In a discussion on model evaluation, alexgshaw argues companies should build their own internal benchmarks rather than relying on public leaderboards — claiming that's the only approach that encourages broadly good models, since public indices invite benchmark gaming.
More from Models
- Claude's model picker screenshot: Opus 5.5, Sonnet 5.5, Fable 5.2, Haiku 5.5 — BLCNYY · 2026-11-10
- Meta's coding model Muse Ember goes live on OpenRouter for agent sub-task fan-out — alexandr_wang · 2026-11-10
- Meta's Muse Ember coding model is now available in OpenCode — alexandr_wang · 2026-11-10
- Stanford shows Grok fails face identification, misidentifying Prof. Tatsunori Hashimoto — stanfordnlp · 2026-11-10
- Anthropic ships Claude Fable 5.2 & Mythos 5.2: Fermat proof in 30 hours, lower prices — Dr_Singularity · 2026-11-10
- Claude Fable 5.2 autonomously speeds up Quantum Espresso DFT simulations by 36x — simonbatzner · 2026-11-10