← Back to feed

Mistral AI (@MistralAI): Le Chonk cooked on visual grounding. Pinpoints exactly where things are in an image. 42% on Dense 200, ahead of even frontier closed models.. https://t.co/QKafkN1qoi

L4 · DeveloperModels & Releases@MistralAI· 10/12/2026

Important benchmark results for developers working with computer vision and multimodal AI systems.

AI Summary

Mistral AI's 'Le Chonk' model achieves 42% on Dense 200 visual , outperforming even frontier closed models.

Excerpt

Le Chonk cooked on visual grounding. Pinpoints exactly where things are in an image. 42% on Dense 200, ahead of even frontier closed models.. https://t.co/QKafkN1qoi

Read Original
0 upvotes · 0 downvotes · 1 min read

Related Articles

L4 · DeveloperModels & ReleasesTechCrunch AI
Meta releases its first frontier model, Muse Wonder

Meta released Muse Wonder, its first frontier AI model with 'active memory' capabilities and benchmark scores comparable to GPT-6 Astra and Fable 5.1.

L3 · BuilderModels & ReleasesThe Verge AI
Meta has some Wonder Full new AI models 

Meta launches Muse Wonder and Muse Wonder Full AI models that compete with GPT-6 Astra and Claude Opus 5.5, available for Muse Code subscribers and API access.

L4 · DeveloperModels & ReleasesMicrosoft AI Blog
Introducing Microsoft’s latest custom chip: Maia 300

Microsoft launches Maia 300, a custom AI inference chip designed for Azure deployments, scaling to multi-gigawatt capacity.

L3 · BuilderModels & ReleasesHacker News
Claude Haiku 5.5

Anthropic releases Claude Haiku 5.5, a significantly cheaper and faster small model optimized for high-volume tasks like summarization, classification, and customer support.

L4 · DeveloperModels & ReleasesLast Week in AI
Last Week in AI #346 - 719 math manuscripts, 2 Western open models, 1 more safety resignation

OpenAI published 719 mathematical proofs from an unreleased frontier model while Mistral and Reflection AI launched open-weight models intended to compete with Chinese AI developments.