Pixtral Large
124B open-weights multimodal model built on Mistral Large 2
About
Pixtral Large, released November 18, 2024, is Mistral's 124B multimodal model: a 123B decoder plus a 1B vision encoder with a 128K context window that fits around 30 high-resolution images. Mistral reports 69.4% on MathVista, ahead of GPT-4o and Claude 3.5 Sonnet on visual math, plus strong DocVQA (94.2%) and ChartQA (88.6%) for document and chart work. The licensing is split: the Mistral Research License covers academic use while commercial use needs the Mistral Commercial License, so it is open-weights but not open-source. At 124B it needs serious GPU memory to self-host.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
MathVista 69.4%, DocVQA 94.2%, ChartQA 88.6% (Mistral-reported)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: Release 2024-11-18, 124B (123B decoder + 1B vision encoder), 128K context verified via Mistral announcement mirrors; MathVista 69.4%, DocVQA 94.2%, ChartQA 88.6% are Mistral-reported; labeled as such; Mistral Research License for research + separate commercial license; not open-source; caveat captured.