Mistral Small 3.2 24B
Mistral AI · released Jun 19, 2025 · mistralai/Mistral-Small-3.2-24B-Instruct-2506
- Type
- Open weightsApache License 2.0
- Params
- 24B
- Context
- 256K
about 192K words of context
Our take
Written Aug 4, 2026Mistral Small is a 24-billion-parameter text-and-image model released in 2025 with a permissive Apache licence. Its input rate is among the lowest of capable mid-size models, making it a good fit for budget production workloads that do not need the latest release.
Pick this for budget production workloads where low input cost matters. Use it for self-hosting on 24GB-class GPUs with a permissive licence, or vision-input tasks at minimal cost. Skip it if you want measured quality scores or the newest 2026 models.
The case for it
- Lowest-cost hosted inference in the 24–27-billion-parameter class.
- Apache 2.0 licence allows unrestricted commercial use and fine-tuning.
The case against it
- Released in 2025, a generation behind current mid-size releases.
- No benchmark scores yet, so there is no measured quality data.
- Only five tracked offers, a thinner hosting market than peers.
How good is it?
We hold no score for this model.
So there is no figure here for everyday use, coding, agent work or writing. That is a gap in our data, not a low score.
Can you run it yourself?
Comfortable fit
GeForce RTX 4090 · 24 GB
Room to spare. 5.7 GB spare means a 10% error in the size would not change the answer.
GeForce RTX 5090 · 32 GB
Room to spare. 13.7 GB spare means a 10% error in the size would not change the answer.
Apple M2 (10-core GPU) · 24 GB
Borderline fit on an estimated size. It leaves 0.9 GB spare on a size we calculated rather than measured, and a 10% error either way would change the answer.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first.
Check against your own machine → · Where to rent it hosted →
Or rent it from someone else
Prices checked 2 hours ago — each listing carries its own date.
The only listing at 256K of context — the other 4 in the table below are not like-for-like. One cheaper row there is outside that comparison: a different context length.
- per 1M tokens
- $0.094 in / $0.25 out
- Context served
- 256K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| DeepInfrafp8Direct and through OpenRouter | $0.075 / $0.20checked 2 hours ago | 128K16K max reply through OpenRouter | 23 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| OpenRouterOpenRouter's own listing | $0.094 / $0.25checked 2 hours ago | 256K | not measured | Unknown | Unknown | Unknown |
| Venice AIfp8Through OpenRouter | $0.094 / $0.25checked 2 hours ago | 256K16K max reply | 27 tok/s | No | No | Confirmed |
| Mistral AIeuThrough OpenRouter | $0.10 / $0.30checked 2 hours ago | 33K26K max reply | 29 tok/s | No | Yes30 days | Confirmed |
| Parasailbf16Through OpenRouter | $0.090 / $0.30checked 2 hours ago | 131K33K max reply | 16 tok/s | No | No | Confirmed |
DeepInfra
Mistral AIAcross the 5 listings we hold: 4 say they do not train on prompts (1 of them only through OpenRouter), 0 say they do and 1 does not say. 4 appear in the zero-retention registry we check (1 of them only through OpenRouter); the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| DeepInfrafp8Direct and through OpenRouter | ✓ | ✓ | ✓ |
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| Venice AIfp8Through OpenRouter | ✓ | ✓ | ✓ |
| Mistral AIeuThrough OpenRouter | ✓ | ✓ | ✓ |
| Parasailbf16Through OpenRouter | ✗ | ✓ | ✓ |
Tool calling: 4 of 5 listings say yes, 1 says no. JSON output: 5 of 5 listings say yes. Strict schema: 5 of 5 listings say yes.
Models people weigh against Mistral Small 3.2 24B
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- No independent board has scored it, so we hold no quality figures at all.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 5 listings does not say whether it trains on prompts, and 1 answers only through OpenRouter, not for its own listing.
- We hold no batch or off-peak rate for any of its listings.
- We hold a decode speed for it, but no prompt-processing (prefill) figure, so how long the input side of a job takes is unknown to us.
Licence and identifiers
What the licence allowsApache License 2.0, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Apache License 2.0
Fully permissive: commercial use, redistribution, and derivatives allowed. Requires attribution and a copy of the license. Includes an express patent grant.
Identifiers
- Hugging Face
- mistralai/Mistral-Small-3.2-24B-Instruct-2506
- Architecture
- Dense
- Takes in, gives back
- Text and images in, text out
- Catalogue slug
- mistralai-mistral-small-3-2-24b