Pixtral 12B
Open Weights
Mistral AI
DISCONTINUED. Mistral's first multimodal model (September 2024), a 12B open-weight model with a dedicated vision encoder that understands images alongside text at native resolution. Released under Apache 2.0, it brought strong open multimodal performance to consumer-scale hardware. Mistral's own launch page now carries a deprecation banner stating the model is no longer maintained; the hosted version was deprecated on December 2, 2025 with retirement on December 31, 2025.
Strengths
- Open-weight 12B multimodal under Apache 2.0
- Understands images at native resolution
- 128K context window
- Strong open multimodal performance
- Runs on a single GPU
Caveats
- DISCONTINUED as a hosted API - Mistral deprecated it and no longer maintains it
- Mistral directs users to its current vision and multimodal models
- 12B ceiling below larger multimodal models
- No audio input
- Superseded by Pixtral Large for peak quality
- Vendor
- Mistral AI
- Context Window
- 131,072 tokens
- License
- Apache-2.0
- Release Date
- 2024-09-17
- Modalities
- textimage
Capabilities
Vision
Audio
Video
Tool Use
Resources
Pricing
Open weights under Apache 2.0 - free to self-host. The hosted API model pixtral-12b-2409 was deprecated 12/2/2025 and retired 12/31/2025 (replacement: Ministral 3 14B).