Mistral Large 4 trails GLM-5.3 on Artificial Analysis index, fueling 'eval crisis' talk
Yuchenj_UW · x · 2026-10-06
Yuchen Jin points out that Mistral's newly released Mistral Large 4 (aka Le Chonk) ranks well behind GLM-5.3 on the Artificial Analysis Intelligence Index — even behind GLM-5.3 Flash. He quips that we're in an "eval crisis": benchmark results diverge so wildly from vendor claims that judging real model capability is getting harder.
Related event: Mistral Large 4 Review: France Back in Third Place, but Value Questioned(11 posts)→
More from Fun
- 2023 HN thread trashing Anthropic resurfaces as the ultimate 'little did they know' moment — Sea-Ad-7031 · 2026-11-21
- 'Call me when Claude brings back the dodo': AI circles shrug at Riemann proof — jxmnop · 2026-11-21
- 940-page claimed Riemann proof scans as 100% human-written across 594,184 words — burny_tech · 2026-11-21
- Bio-inspired synth "Bloom" grows a simulated plant that plays unique music for days, built with Fable 5.2 + Opus 5.5 — ssh4net · 2026-11-21
- OpenAI employees joke they hold no insider answer to whether P=NP — sjgadler · 2026-11-21
- The Martian's chemistry error: one liter of oxygen can't yield two liters of water — xuanalogue · 2026-11-21