Contamination MapText / LLM / Model
Ministral 3
- Organisation
- Mistral AI
- Family
- Mistral · version 2512
- Released
- 2025-12-02 (Mistral AI blog 'Introducing Mistral 3' dated December 2, 2025; technical report arXiv v1 submitted 2026-01-13)
- Checkpoints
- mistralai/Ministral-3-3B-Base-2512, mistralai/Ministral-3-3B-Instruct-2512, mistralai/Ministral-3-3B-Reasoning-2512, mistralai/Ministral-3-8B-Base-2512, mistralai/Ministral-3-8B-Instruct-2512, mistralai/Ministral-3-8B-Reasoning-2512, mistralai/Ministral-3-14B-Base-2512, mistralai/Ministral-3-14B-Instruct-2512, mistralai/Ministral-3-14B-Reasoning-2512
- 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
- ?Unknown15
“After weight initialization, each child model is trained on a mixture of text-only and interleaved text with image data with logit distillation from a teacher model.”
Other declared training data
-
mixture of text-only and interleaved text with image data (not yet matched to a dataset) · split: unknown
Sources not described; 1-3T tokens per the paper.
Evidence
“After weight initialization, each child model is trained on a mixture of text-only and interleaved text with image data with logit distillation from a teacher model.”
arxiv.org · paper · Section 3.1 Pretraining, Distillation · retrieved 2026-10-15 -
Mistral Small 3.1 (parent model; pruning initialisation and distillation teacher) (not yet matched to a dataset) · split: unknown
Ministral 3 models are pruned from and distilled from Mistral Small 3.1, whose own pre-training data is undisclosed.
Evidence
“with the distillation teacher being Mistral Small 3.1 for all variants”
arxiv.org · paper · Section 3.1 Pretraining · retrieved 2026-10-15 -
curated multimodal and text-only instruction-following data (not yet matched to a dataset) · split: unknown · fine-tuning
Evidence
“using a curated dataset comprising high-quality multimodal and text-only instruction-following data.”
arxiv.org · paper · Section 3.2 Post-Training: Ministral Instruct · retrieved 2026-10-15 -
reasoning traces (short and long chain-of-thought) (not yet matched to a dataset) · split: unknown · fine-tuning
Reasoning variants only.
Evidence
“The reasoning traces come from a diverse set of domains including mathematics, coding, general dialogue, instruction following, multilingual tasks, tool use, and visual reasoning.”
arxiv.org · paper · Section 3.3.1 Reasoning Supervised Fine-Tuning · 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
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
The shipped SYSTEM_PROMPT.txt says 'Your knowledge base was last updated on 2023-10-01.'; not recorded as training_cutoff (same boilerplate as older Mistral prompts).
Sources
- Ministral 3 (arXiv v1) · paper · retrieved 2026-10-15
- Ministral-3-14B-Base-2512 model card · model card · retrieved 2026-10-15
- Ministral-3-14B-Instruct-2512 model card · model card · retrieved 2026-10-15
- Introducing Mistral 3 (Mistral AI blog) · website · retrieved 2026-10-15
- Ministral 3 SYSTEM_PROMPT.txt · code · retrieved 2026-10-15