Pixtral 12B

TL;DR

Pixtral-12B is a versatile 12-billion-parameter multimodal language model that excels in understanding images and documents, outperforming larger models and maintaining strong natural language capabilities.

Contribution

Introduces Pixtral-12B with a new vision encoder, long context processing, and an open-source benchmark for multimodal evaluation, advancing open-source multimodal LLMs.

Findings

Outperforms similar-sized open models like Llama-3.2 11B and Qwen-2-VL 7B

Surpasses larger models such as Llama-3.2 90B in multimodal tasks

Supports processing images at natural resolution and arbitrary number of images in long contexts

Abstract

We introduce Pixtral-12B, a 12--billion-parameter multimodal language model. Pixtral-12B is trained to understand both natural images and documents, achieving leading performance on various multimodal benchmarks, surpassing a number of larger models. Unlike many open-source models, Pixtral is also a cutting-edge text model for its size, and does not compromise on natural language performance to excel in multimodal tasks. Pixtral uses a new vision encoder trained from scratch, which allows it to ingest images at their natural resolution and aspect ratio. This gives users flexibility on the number of tokens used to process an image. Pixtral is also able to process any number of images in its long context window of 128K tokens. Pixtral 12B substanially outperforms other open models of similar sizes (Llama-3.2 11B \& Qwen-2-VL 7B). It also outperforms much larger open models like Llama-3.2 90B while being 7x smaller. We further contribute an open-source benchmark, MM-MT-Bench, for evaluating vision-language models in practical scenarios, and provide detailed analysis and code for standardized evaluation protocols for multimodal LLMs. Pixtral-12B is released under Apache 2.0 license.