Mistral Small 3.2
Mistral Small 3.2 is Mistral AI's minor update to Small 3.1, improving instruction following, cutting repetitive generations, and refining function calling.
Model Overview
Capabilities, design details, and architectural traits
Mistral Small 3.2 - A Focused Minor Update to Small 3.1
Mistral Small 3.2 is documented as a minor update to Mistral Small 3.1, built around a narrow, named set of fixes rather than a broad capability overhaul. Mistral AI describes it as improving in three specific categories while matching or slightly improving everywhere else.
| Trait | Documented Behavior |
|---|---|
| Sharper Instruction Following | Follows precise instructions, including strict formatting requests, more closely than Small 3.1. |
| Reduced Repetition | Cuts infinite and repetitive generations roughly in half on long, challenging prompts compared to Small 3.1. |
| More Robust Function Calling | Uses an improved function-calling template for more reliable tool-use behavior than Small 3.1. |
Everything Else Stays the Same
Mistral AI states that Small 3.2's key features are the same as Small 3.1's: the same vision support, the same 128k context window, and the same Apache 2.0 license. This is a model explicitly scoped as a targeted correction rather than a new generation, distinguishing it from typical version bumps that rewrite the whole feature list.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Mistral | Anthropic | Anthropic |
| Release Date | June 20, 2025 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | Oct 2023 | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 128K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.07 Best Input Pricing | $4 | $2 |
| Output Pricing | $0.20 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | imagetext | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 8.2 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 12.5 | - | - |
| Agentic Index | 1.0 | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from Mistral
Other models by Mistral
Mistral Large 4 Preview
Mistral Medium 3.5
Mistral Large 3
Mistral Medium 3.1
Top AI Models
Leading alternatives by intelligence score
