Pixtral 12B 2409 EU AI Act Compliance Profile

Model Card for Pixtral-12B-2409

The Pixtral-12B-2409 is a Multimodal Model of 12B parameters plus a 400M parameter vision encoder.

For more details about this model please refer to our release blog post.

Feel free to try it here

Key features

• Natively multimodal, trained with interleaved image and text data
• 12B parameter Multimodal Decoder + 400M parameter Vision Encoder
• Supports variable image sizes
• Leading performance in its weight class on multimodal tasks
• Maintains state-of-the-art performance on text-only benchmarks
• Sequence length: 128k
• License: Apache 2.0

Benchmarks

The performance of Pixtral-12B-2409 compared to multimodal models.
All models were re-evaluated and benchmarked through the same evaluation pipeline.

Multimodal Benchmarks

| | Pixtral 12B | Qwen2 7B VL | LLaVA-OV 7B | Phi-3 Vision | Phi-3.5 Vision | |:-------------------:|:-------------:|:----------:|:-------------:|:--------------:|:--------------:| | MMMU (CoT) | <ins>52.5</ins> | 47.6 | 45.1 | 40.3 | 38.3 | | Mathvista (CoT) | <ins>58.0</ins> | 54.4 | 36.1 | 36.4 | 39.3 | | ChartQA (CoT) | <ins>81.8</ins> | 38.6 | 67.1 | 72.0 | 67.7 | | DocVQA (ANLS) | 90.7 | <ins>94.5</ins> | 90.5 | 84.9 | 74.4 | | VQAv2 (VQA Match) | <ins>78.6</ins> | 75.9 | 78.3 | 42.4 | 56.1 |

Instruction Following

| | Pixtral 12B | Qwen2 7B VL | LLaVA-OV 7B | Phi-3 Vision | Phi-3.5 Vision | |:-------------------:|:-------------:|:----------:|:-------------:|:--------------:|:--------------:| | MM MT-Bench | <ins>6.05</ins> | 5.43 | 4.12 | 3.70 |4.46 | | Text MT-Bench | <ins>7.68</ins>

Your obligations as a Pixtral 12B 2409 deployer

Track your progress

Sign up to track obligations, receive deadline reminders, and generate audit packages.

Tags

Languages

Library

Base Model