Pixtral Large
Open Weights
Mistral AI
DISCONTINUED. Mistral's 124B multimodal model (November 2024), built on Mistral Large 2 with an integrated vision encoder for frontier-class image understanding. It handles documents, charts, and natural images alongside text with a 128K context window. Mistral's launch page is now titled [Deprecated] Pixtral Large and states the model is no longer maintained; the hosted version was deprecated on February 27, 2026.
Strengths
- 124B multimodal model with frontier-class image understanding
- Built on Mistral Large 2 with an integrated vision encoder
- Handles documents, charts, and natural images
- 128K context window
- Strong combined text + vision reasoning
Caveats
- DISCONTINUED as a hosted API - Mistral deprecated it and no longer maintains it
- Mistral directs users to its current vision and multimodal models
- Research license for weights (commercial license required for production)
- 124B - demanding to self-host at full precision
- No audio input
- Vendor
- Mistral AI
- Context Window
- 131,072 tokens
- License
- Mistral Research License
- Release Date
- 2024-11-18
- Modalities
- textimage
Capabilities
Vision
Audio
Video
Tool Use
Resources
Pricing
Weights available under the Mistral Research License (commercial license available); 124B (123B multimodal decoder + 1B vision encoder). No longer served on the Mistral API: pixtral-large-2411 was deprecated 2/27/2026 and retired 5/31/2026 (replacement: Mistral Medium 3.5).