Contamination MapText / LLM / Model
Pixtral 12B
- Organisation
- Mistral AI
- Family
- Pixtral · version 2409
- Released
- 2024-09-17 (Mistral blog post 'Pixtral 12B' dated September 17, 2024 (https://mistral.ai/news/pixtral-12b); HF repo created 2024-09-11)
- Checkpoints
- mistralai/Pixtral-12B-2409, mistralai/Pixtral-12B-Base-2409
- Weights
- Open · Apache 2.0
- Links
- PaperModel card
Training data disclosure: Vague
Level determined for 0% of benchmark cells.
- TestTest data seen0
- SeenSeen, split unclear0
- TrainTraining data seen0
- Related data seen0
- NoNot seen0
- ?Unknown22
“Pixtral is an instruction tuned model which is pretrained on large scale interleaved image and text documents”
“Natively multimodal, trained with interleaved image and text data”
Other declared training data
-
pre-training data of base model Mistral NeMo 12B (broad web data: may contain benchmark test sets) · split: unknown
Mistral NeMo is not in the registry; its training data are undisclosed.
Evidence
“Pixtral 12B is built on top of Mistral Nemo 12B [15]”
arxiv.org · paper · Section 2.1 Multimodal Decoder · retrieved 2026-10-15 -
large-scale interleaved image and text documents (broad web data: may contain benchmark test sets) · split: unknown
Also used for the vision encoder (Pixtral-ViT, trained from scratch); no source named.
Evidence
“Pixtral is an instruction tuned model which is pretrained on large scale interleaved image and text documents”
arxiv.org · paper · Section 1 · retrieved 2026-10-15
Benchmark cells
Classic LLM benchmarks
Unknown MMLU (Measuring Massive Multitask Language Understanding) · training data not disclosed in enough detail
Unknown GSM8K (Grade School Math 8K) · training data not disclosed in enough detail
Unknown MATH (Mathematics Aptitude Test of Heuristics) · training data not disclosed in enough detail
Unknown HumanEval · training data not disclosed in enough detail
Unknown ARC-Challenge (AI2 Reasoning Challenge, Challenge Set) · training data not disclosed in enough detail
Unknown HellaSwag · training data not disclosed in enough detail
Unknown TruthfulQA · training data not disclosed in enough detail
Unknown WinoGrande · training data not disclosed in enough detail
Multimodal LLM benchmarks
Unknown MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning) · training data not disclosed in enough detail
Unknown MMMU-Pro · training data not disclosed in enough detail
Unknown MathVista · training data not disclosed in enough detail
Unknown ChartQA · training data not disclosed in enough detail
Unknown DocVQA · training data not disclosed in enough detail
Unknown AI2D (AI2 Diagrams) · training data not disclosed in enough detail
Unknown CharXiv · training data not disclosed in enough detail
Recent LLM benchmarks
Unknown GPQA Diamond · training data not disclosed in enough detail
Unknown MMLU-Pro · training data not disclosed in enough detail
Unknown LiveCodeBench (code generation, release_v6) · training data not disclosed in enough detail
Unknown SWE-bench Verified · training data not disclosed in enough detail
Unknown AIME 2024 · training data not disclosed in enough detail
Unknown AIME 2025 · training data not disclosed in enough detail
Unknown MATH-500 · training data not disclosed in enough detail
Notes
Vision encoder (Pixtral-ViT, 400M) trained from scratch; no training data or instruction-tuning data named, no decontamination statement. Model is marked deprecated on the Mistral website. Accepts image input: “This paper describes Pixtral 12B, a multimodal language model trained to understand both images and text”
Sources
- Pixtral 12B (arXiv 2410.07073v2) · paper · retrieved 2026-10-15
- Pixtral-12B-2409 model card · model card · retrieved 2026-10-15
- Mistral blog: Pixtral 12B (2024-09-17) · website · retrieved 2026-10-15