Mistral Small 3.1
Mistral Small 3.1 is Mistral AI's vision upgrade to the text-only Small 3, adding image understanding, a Tekken tokenizer, and a 128k context window.
Model Overview
Capabilities, design details, and architectural traits
Mistral Small 3.1 - Vision Added to Small 3 Without Losing Text Performance
Mistral Small 3.1 builds on the text-only Mistral Small 3 (2501) by adding vision understanding and a longer context window, while stating that text performance is not compromised in the process.
| Trait | Documented Behavior |
|---|---|
| Text-to-Vision Upgrade | Adds vision understanding on top of the previously text-only Mistral Small 3, without reducing text performance. |
| Tekken Tokenizer | Uses a Tekken tokenizer with a 131k-token vocabulary. |
| 128k Context Extension | Expands the context window up to 128k tokens, longer than the prior Small 3 release. |
Released as Base and Instruct Together
Mistral AI released both Small 3.1 Base and Instruct checkpoints at the same time, explicitly to let the community build downstream reasoning models on top of the base weights rather than only shipping a closed instruction-tuned version.
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 | March 17, 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 | $0.08 Best Input Pricing | $4 | $2 |
| Output Pricing | $0.14 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | textimage | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 7.1 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 26.3 | - | - |
| Agentic Index | 1.2 | - | - |
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
