Pixtral-12B

Unknown Developer

This powerful multimodal model uses 12 billion parameters, including a 400 million parameter vision encoder, to expertly interpret both images and documents. It delivers exceptional results on tasks requiring combined understanding of text and images, without sacrificing strong performance in text-only applications. The model is also flexible, accommodating varying image sizes and processing multiple images simultaneously for richer contextual awareness.

Model Specifications

Technical details and capabilities of Pixtral-12B

Core Specifications

12.4B Parameters

Model size and complexity

128.0K / 8.2K

Input / Output tokens

September 16, 2024

Release date

Capabilities & License

Multimodal Support
Supported
Web Hydrated
No
License
apache-2.0

Resources

API Reference
https://docs.mistral.ai/platform/endpoints/
Playground
https://chat.mistral.ai
Code Repository
https://huggingface.co/mistralai/Pixtral-12B-2409

Performance Insights

Check out how Pixtral-12B handles various AI tasks through comprehensive benchmark results.

100
75
50
25
0
90.7
DocVQA
90.7
(91%)
81.8
ChartQA
81.8
(82%)
78.6
VQAv2
78.6
(79%)
76.8
MT-Bench
76.8
(77%)
72
HumanEval
72
(72%)
69.2
MMLU
69.2
(69%)
61.3
IFEval
61.3
(61%)
60.5
MM MT-Bench
60.5
(61%)
58.0
Mathvista
58.0
(58%)
52.7
MM IF-Eval
52.7
(53%)
52.5
MMMU
52.5
(53%)
48.1
Math
48.1
(48%)
DocVQA
ChartQA
VQAv2
MT-Bench
HumanEval
MMLU
IFEval
MM MT-Bench
Mathvista
MM IF-Eval
MMMU
Math

Detailed Benchmarks

Dive deeper into Pixtral-12B's performance across specific task categories. Expand each section to see detailed metrics and comparisons.

Coding

HumanEval

Current model
Other models
Avg (70.0%)

Non categorized

Mathvista

Current model
Other models
Avg (55.4%)

ChartQA

Current model
Other models
Avg (83.6%)

DocVQA

Current model
Other models
Avg (90.9%)

VQAv2

Current model
Other models
Avg (79.2%)

IFEval

Current model
Other models
Avg (75.4%)

Providers Pricing Coming Soon

We're working on gathering comprehensive pricing data from all major providers for Pixtral-12B. Compare costs across platforms to find the best pricing for your use case.

OpenAI
Anthropic
Google
Mistral AI
Cohere

Share your feedback

Hi, I'm Charlie Palars, the founder of Deepranking.ai. I'm always looking for ways to improve the site and make it more useful for you. You can write me through this form or directly through X at @palarsio.

Your feedback helps us improve our service

Stay Ahead with AI Updates

Get insights on Gemini Pro 2.5, Sonnet 3.7 and more top AI models