mistralai
Mistral: Mistral Small 3.2 24B
mistralai/mistral-small-3.2-24b-instruct
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
mistralai/mistral-small-3.2-24b-instruct is the OpenRouter listing for Mistral's 24B-parameter Mistral Small 3.2 Instruct (2506), a text-and-image input, text-output model with a very large context window and tool/function-calling support. Mistral describes it as a minor update to Small 3.1 focused on instruction following, fewer repetition or infinite-generation errors, and more robust function calling. Its published results suggest meaningful chat and instruction gains with broadly similar STEM performance and mixed vision results, so it should be treated as an incremental rather than a guaranteed improvement.[1][2]
Strengths
- Accepts text and image inputs and returns text, enabling vision-aware assistant and document-style tasks.[1][2]
- Mistral reports substantially better instruction following and chat performance than Small 3.1, with developer benchmarks also indicating fewer infinite or repetitive generations.[2]
- The catalog and model documentation list structured outputs and broad tool controls; Mistral describes function calling as more robust and demonstrates it through vLLM.[1][2]
Limitations
- The evidence here is a catalog record plus the developer model card; no independent evaluation, deployment measurements, or workload-specific validation is provided.[1][2]
- Developer benchmarks indicate vision results are mixed relative to Small 3.1 rather than uniformly improved.[2]
- Self-hosting requires substantial GPU memory in bf16 or fp16 and a comparatively involved recommended serving setup.[2]
Where to use it
- A reasonable candidate for assistants that must follow precise instructions, especially where repetitive or endless responses were a problem.[2]
- Potentially useful for image-grounded chat or document and chart question answering, but validate accuracy on your own inputs because reported vision results are mixed.[1][2]
- Worth testing for tool-using workflows or structured API outputs using the documented vLLM function-calling path.[1][2]
Sources
- [1] OpenRouter catalog · Mistral: Mistral Small 3.2 24BSources checked: Nov 13, 2026
- [2] Model documentation · Mistral: Mistral Small 3.2 24BSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Mistral-Small-3.2-24B-Instruct-2506 is an updated 24B parameter model from Mistral optimized for instruction following, repetition reduction, and improved function calling. Compared to the 3.1 release, version 3.2 significantly improves accuracy on...
Specifications
- Model ID
- mistralai/mistral-small-3.2-24b-instruct
- Canonical ID
- mistralai/mistral-small-3.2-24b-instruct-2506
- Creator
- mistralai
- Added to OpenRouter
- 20.06.2025
- Context length
- 256,000
- Modality
- text+image->text
- Input modalities
- image, text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.100 / out $0.300
- Knowledge cutoff
- 31.10.2023
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | — |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | — |
| Document analysis | No measurement | — |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Design Arena | models · codecategorieswin rate 39.8% · rank 141 | 897Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · datavizwin rate 43.3% · rank 128 | 929Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · gamedevwin rate 39.4% · rank 141 | 879Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 40.5% · rank 135 | 898Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · websitewin rate 38.3% · rank 144 | 888Elo | CC — below the B band | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| DeepInfrafp8 | 128,000 | $0.075 | $0.200 | — | — | 100% |
| Parasailbf16 | 131,072 | $0.090 | $0.300 | — | — | 58.62% |
| Venicefp8 | 256,000 | $0.094 | $0.250 | — | — | 100% |
| Mistral | 32,768 | $0.100 | $0.300 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.