Pixtral Large
Pixtral Large is Mistral’s larger vision-language model, combining a 123B language decoder with a 1B vision encoder. It processes interleaved text and images for document interpretation, chart analysis, and visual reasoning, with a 128K context window. Downloadable weights are offered under the Mistral Research License, with separate commercial licensing required. The model is derived from Mistral Large 2 and complements the earlier 12B Pixtral. Visual grounding and extracted text can still be inaccurate on complex documents.
2024-11-18
124B (123B decoder + 1B vision encoder)
Decoder-only Transformer with vision encoder
Mistral Research License; commercial license separately available
Specifications
- Parameters
- 124B (123B decoder + 1B vision encoder)
- Architecture
- Decoder-only Transformer with vision encoder
- License
- Mistral Research License; commercial license separately available
- Context Window
- 131,072 tokens
- Type
- multimodal
- Modalities
- textimage
Benchmark Scores
Advanced Specifications
- Model Family
- Pixtral
- Finetuned From
- Mistral Large 2
- API Access
- Available
- Chat Interface
- Not Available
- Multilingual Support
- Yes
Capabilities & Limitations
- Capabilities
- visiondocument understandingchart analysismultilingual
- Known Limitations
- Research weights have commercial licensing restrictionsVisual interpretation may be inaccurate
- Notable Use Cases
- document question answeringimage reasoningenterprise visual assistants