Pixtral was Mistral AI's line of models that understand images as well as text; Mistral has since retired it in favour of newer multimodal models.
In plain words
Pixtral 12B, released in September 2024, could read pictures at their original size and shape rather than squashing them, and could take many images at once within a 128K-token context. Mistral said it did this without getting worse at plain text tasks. A larger Pixtral Large followed in November 2024. Both have been retired from Mistral's API, and image understanding now comes built into models such as Mistral Small 4 and Ministral 3.
A quick example
Imagine giving Pixtral a scanned receipt and a photo of a bar chart in one message and asking for the total spent and the tallest bar. It reads both images alongside your question.
Where you'll meet it
- Reading charts, diagrams and documents, before retirement.
Related
Sources
- DocsPixtral 12B, Mistral AI · read 29 Sept 2026
- DocsModels overview, Mistral AI docs · read 29 Sept 2026
- DocsIntroducing Mistral Small 4, Mistral AI · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.