The latest iteration of Mistral Small, version 3.2 (2506) brings enhanced performance and capabilities. With 24 billion parameters, this model delivers state-of-the-art results across text generation tasks with improved efficiency.
Added Apr 15, 2025
Model weightsContext Window
128.0K
Max Output
131.1K
Avg output tokens (7d)
536 tokens
Input Price (Auto)
$0.20/1M
Output Price (Auto)
$0.40/1M
Cache Read (Auto)
$0.10/1M
Benchmarks
Benchmarks
Performance metrics and benchmarks
Sourced from Artificial Analysis.
Intelligence Index
7.0
Coding Index
12.5
Agentic work
AutomationBench-AA
Workflow automation with guardrail penalties
0.3%
Better than 7% of models compared
AA-Briefcase
Agentic knowledge work (Elo)
0 Elo
Better than 2% of models compared
GDPval-AA v2
Economically valuable tasks (Elo)
29 Elo
Better than 5% of models compared
Document reasoning
GDP.pdf
Professional PDF reasoning: all-pass rate
0.0%
Better than 1% of models compared
AA-LCR v1.1
Long context reasoning with updated grading
20.3%
Better than 24% of models compared
Reasoning
HLE
Humanity's Last Exam
4.3%
Better than 19% of models compared
IFBench
Instruction-following benchmark
33.5%
Better than 23% of models compared
Coding
Terminal-Bench v4.0
Practical coding and terminal tasks
0.0%
Better than 18% of models compared
SciCode
Python programming for scientific computing
28.6%
Better than 8% of models compared
LiveCodeBench
Contamination-free coding benchmark
27.5%
Better than 30% of models compared
Math
AIME 2025
American Invitational Mathematics Examination 2025
27.0%
Better than 29% of models compared
AIME
American Invitational Mathematics Examination
32.3%
Better than 59% of models compared
Math-500
Diverse mathematical problem solving benchmark
88.3%
Better than 59% of models compared
Knowledge
MMLU-Pro
Professional and academic subject knowledge
68.1%
Better than 32% of models compared
Legacy benchmarks
GPQA Diamond (legacy)
Graduate-level scientific reasoning
50.5%
Better than 24% of models compared
Terminal-Bench Hard (legacy)
Agentic coding and terminal use
6.8%
Better than 37% of models compared
T²-Bench Telecom (legacy)
Conversational AI agents in dual-control scenarios
29.5%
Better than 37% of models compared
AA-LCR (unversioned / legacy)
Long context reasoning evaluation
20.3%
Better than 24% of models compared
Last updated Sep 11, 2026
Artificial AnalysisProviders
Auto routing is available for this model. Explicit provider selection is not available.
Loading provider options…
Related text models
Compare Mistral Small 3.2 24B (2506) with similar models from the same provider or model family.
Mistral Small 3 24B (2501)
mistralai/mistral-small-24b-instruct-2501Mistral Small 3 24B (2501) hosted by IONOS in Berlin, Germany. Zero data retention.
Mistral Small 3.1 24B (2503)
mistralai/mistral-small-3.1-24b-instructBuilding upon Mistral Small 3 (2501), Mistral Small 3.1 (2503) adds state-of-the-art vision understanding and enhances long context capabilities up to 128k tokens without compromising text performance. With 24 billion parameters, this model achieves top-tier capabilities in both text and vision tasks.
Mistral Small 4 119B Thinking
mistralai/mistral-small-4-119b-2603:thinkingMistral Small 4 with reasoning enabled (reasoning_effort=high). A hybrid MoE model with deep step-by-step reasoning for complex prompts, coding, and multi-step problem solving.
Mistral Small 4 119B
mistralai/mistral-small-4-119b-2603Mistral Small 4 is a hybrid MoE model that unifies instruct, reasoning, and coding behavior in a single multimodal model. It supports text and image input, native function calling, JSON output, and per-request reasoning effort controls.
Mistral Large 3 675B
mistralai/mistral-large-3-675b-instruct-2512Mistral Large 3 675B is Mistral AI's flagship language model featuring advanced rope scaling and Eagle speculative decoding. Delivers exceptional performance across reasoning, coding, and multilingual tasks.
Mistral Devstral Small 2505
mistralai/devstral-small-2505OpenHands+Devstral is 100% local 100% open, and is SOTA for the category on SWE-Bench Verified: 46.8% accuracy.