Mistral's Le Chonk scores 42% on Dense 200 visual grounding benchmark
Trending storyDeveloping
Mistral's Le Chonk scores 42% on Dense 200 visual grounding benchmark
1 article1 sourcesince Oct 12Last article 1h ago ·
Overview
AISummary of 1 article
Mistral AI says its Le Chonk model scores 42% on the Dense 200 visual grounding benchmark, which tests how precisely a model locates objects in an image.
In a post on X, the company says the score is ahead of frontier closed models. The post gives no comparison figures for those models, and no independent evaluation of the result has been reported.
Written by AI from the articles below · updated Oct 12, 1:05 PM ET
Check the sources:
Latest developmentOct 12, 12:49 PM ET
Mistral's Le Chonk scores 42% on Dense 200 visual grounding benchmarkArticle timeline
The articles in this story. Times are ET.
Oct 12
Mistral AI@MistralAIOfficialMistral's Le Chonk scores 42% on Dense 200 visual grounding benchmarkAIMistral AI says its Le Chonk model pinpoints exactly where objects are in an image. It scores 42% on the Dense 200 benchmark, which the post says is ahead of even frontier closed models.

Heat trend
Not enough continuous observations to show a trend yet.