Skip to content

Pixtral-12B-2409

A 12B multimodal model with a 400M vision encoder

Visit official site

What Pixtral-12B-2409 is

The model is described as a Mistral AI multimodal release combining a 12B multimodal decoder with a 400M vision encoder for image and text tasks.

What you can do with it

  • Work with interleaved image and text data
  • Accept images of different sizes
  • Run long sequences up to 128k

Getting started

  • Install vLLM and mistral_common
  • Create an instance with the model name and sampling settings
  • Prepare a text prompt and an image URL

Pros & cons

✓ What we like

  • A sequence length of 128k is listed
  • Image and text inputs are handled together

! What to watch out for

  • Setup asks for vLLM and mistral_common libraries
  • Input preparation needs a text prompt and an image URL

FAQ

How large are the components?

The material names a 12B decoder and a 400M vision encoder.

What license applies?

The material states the Apache 2.0 license.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review