# Mistral Small 3.1

> A 24B open-weight model that adds vision and a 128K context to Mistral's Small line.

- Maker: Mistral AI
- Line: Mistral Small
- Canonical page: https://ai-tldr.dev/models/mistral-small-3-1/
- Updated: 2026-06-18

## In short

Mistral Small 3.1 is a 24-billion-parameter open-weight model released by Mistral AI on March 17, 2025, under the Apache 2.0 license. It has a 128K-token context window and costs $0.351 per million input tokens and $0.555 per million output tokens. You can use it through Mistral AI (La Plateforme), OpenRouter and Ollama.

## Facts

| Fact | Value | Source |
| --- | --- | --- |
| Context window | 128K | [1] |
| Input price | $0.351 / 1M tokens | [2] |
| Output price | $0.555 / 1M tokens | [2] |
| Licence | Apache 2.0 · open weights | [1] |
| Released | 17 March 2025 | [3] |
| Parameters | 24B | [1] |
| Knowledge cutoff | October 2023 | [1] |

## Sources

1. [Mistral-Small-3.1-24B-Instruct-2503 — Hugging Face model card](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503)
2. [Mistral Small 3.1 — OpenRouter (API pricing)](https://openrouter.ai/mistralai/mistral-small-3.1-24b-instruct)
3. [Mistral Small 3.1 — official announcement](https://mistral.ai/news/mistral-small-3-1/)

## Overview

Mistral Small 3.1 is a 24-billion-parameter open-weight model released by Mistral AI on March 17, 2025, under the Apache 2.0 license. Carrying the API name mistral-small-2503, it upgrades the earlier Mistral Small 3 by adding state-of-the-art vision understanding and extending the context window to 128K tokens, all without sacrificing the text performance of the previous version.

Mistral positions Mistral Small 3.1 as the best model in its weight class, claiming it outperforms comparable small models such as Google's Gemma 3 and OpenAI's GPT-4o Mini while running at roughly 150 tokens per second. It ships with native function calling and JSON output for agentic use, strong multilingual coverage across dozens of languages, and a compact footprint that Mistral says can run on a single RTX 4090 or a 32GB-RAM Mac once quantized.

Both a base checkpoint and an instruction-tuned checkpoint are published on Hugging Face. Beyond Mistral's own La Plateforme API, Mistral Small 3.1 is available through Google Cloud Vertex AI and a range of third-party hosts including OpenRouter, NVIDIA NIM, and Ollama, making it a practical drop-in for self-hosted and cost-sensitive multimodal workloads.

## Benchmarks

- MMLU: 80.62% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- MMLU Pro (5-shot CoT): 66.76% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- GPQA Diamond (5-shot CoT): 45.96% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- MATH: 69.3% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- HumanEval: 88.41% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- MBPP: 74.71% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- MMMU (vision): 64% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- MathVista: 68.91% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- ChartQA: 86.24% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- DocVQA: 94.08% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- AI2D: 93.72% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- RULER 32K (long context): 93.96% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- RULER 128K (long context): 81.2% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))
- Multilingual (average): 71.18% ([source](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503))

## Pricing

- Input: $0.351 / 1M tokens per 1M tokens
- Output: $0.555 / 1M tokens per 1M tokens

OpenRouter list price for mistral-small-3.1-24b-instruct. As an open-weight Apache-2.0 model it can also be self-hosted at no per-token cost.

Source: https://openrouter.ai/mistralai/mistral-small-3.1-24b-instruct

## Strengths

- Adds vision/image understanding to the 24B Small line while keeping text quality, with strong document and chart scores (DocVQA 94.1%, AI2D 93.7%, ChartQA 86.2%)
- Long 128K context with solid retrieval (RULER 32K 94.0%, RULER 128K 81.2%)
- Apache 2.0 license with both base and instruct weights on Hugging Face — fully self-hostable and fine-tunable
- Compact enough for single-GPU or 32GB-RAM local deployment, with ~150 tokens/sec inference
- Native function calling and JSON output for agentic and tool-using workflows
- Broad multilingual coverage (71.2% average across European, East Asian, and Middle Eastern language sets)

## Best for

- Self-hosted multimodal assistants that need to read images, documents, and charts
- On-device or single-GPU deployment where data privacy or latency rules out cloud APIs
- Document understanding and visual Q&A (forms, receipts, diagrams)
- Agentic pipelines using native function calling and structured JSON output
- Multilingual chat and reasoning across European, East Asian, and Middle Eastern languages
- Domain fine-tuning (legal, medical, technical) on top of the open base checkpoint

## How to access

- [Mistral AI (La Plateforme)](https://docs.mistral.ai/models/mistral-small-3-1-25-03) — model id `mistral-small-2503`
- [OpenRouter](https://openrouter.ai/mistralai/mistral-small-3.1-24b-instruct) — model id `mistralai/mistral-small-3.1-24b-instruct`
- [Ollama](https://ollama.com/library/mistral-small3.1) — model id `mistral-small3.1`

## Mistral Small versions

- [Mistral Small 4](https://ai-tldr.dev/models/mistral-small-4/) (16 March 2026)
- [Mistral Small 3.2](https://ai-tldr.dev/models/mistral-small-3-2/) (20 June 2025)
- Mistral Small 3.1 (17 March 2025) — this page
- [Mistral Small 3](https://ai-tldr.dev/models/mistral-small-3/) (30 January 2025)
- [Mistral Small (24.09)](https://ai-tldr.dev/models/mistral-small-2-409/) (17 September 2024)

## FAQ

### What is Mistral Small 3.1?

Mistral Small 3.1 is a 24-billion-parameter open-weight language model released by Mistral AI on March 17, 2025, under the Apache 2.0 license. Its API name is mistral-small-2503. It adds vision (image) understanding and a 128K-token context window to the earlier Mistral Small 3.

### Is Mistral Small 3.1 open source and free to use?

The weights are released under Apache 2.0 and can be downloaded from Hugging Face in both base and instruction-tuned versions, so you can self-host them for free. If you call it through a hosted API such as Mistral's La Plateforme or OpenRouter, you pay per token instead.

### Does Mistral Small 3.1 support images?

Yes. Unlike the text-only Mistral Small 3, version 3.1 adds vision understanding and scores strongly on visual benchmarks such as DocVQA (94.1%), AI2D (93.7%), and ChartQA (86.2%), making it well suited to reading documents, charts, and diagrams.

### How much does the Mistral Small 3.1 API cost?

On OpenRouter, the mistral-small-3.1-24b-instruct endpoint is listed at about $0.351 per million input tokens and $0.555 per million output tokens. Because it is an open-weight model, self-hosting it avoids per-token API charges entirely.

## Links

- [Mistral Small 3.1 — official announcement](https://mistral.ai/news/mistral-small-3-1/)
- [Mistral Small 3.1 — Mistral docs](https://docs.mistral.ai/models/mistral-small-3-1-25-03)
- [Mistral-Small-3.1-24B-Instruct-2503 — Hugging Face model card](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503)
- [Mistral-Small-3.1-24B-Base-2503 — Hugging Face](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Base-2503)
- [Mistral Small 3.1 — OpenRouter (API pricing)](https://openrouter.ai/mistralai/mistral-small-3.1-24b-instruct)
- [Mistral Small 3.1 — Ollama](https://ollama.com/library/mistral-small3.1)

Source: AI/TLDR LLM registry, https://ai-tldr.dev/models/mistral-small-3-1/