Mistral AI

Mistral Small 3

Generally availableOpen weights33K context

Summary

Mistral Small 3 is an open-weight language model from Mistral AI, released on 30 Jan 2025. It accepts text and generates text, with a 33K-token context window and up to 16K output tokens per response. As of 18 Nov 2026, the lowest listed API price is $0.05 per million input tokens and $0.08 per million output tokens (DeepInfra), across 2 providers we track.

Efficient Mistral model for fast chat, extraction, and production assistants

Last verified 18 Nov 2026. Every figure on this page links to its source.

Specifications

DeveloperMistral AI
Release date30 Jan 2025 [models.dev]
StatusGenerally available [models.dev]
WeightsOpen weights [Hugging Face]
Licenseapache-2.0 [Hugging Face]
Parameters23.6B [Hugging Face]
Context window33K (32,768 tokens) [models.dev]
Max output16K tokens [models.dev]
Knowledge cutoff31 Oct 2023 [models.dev]
InputText [models.dev]
OutputText [models.dev]
Reasoning modeNo [models.dev]
Tool callingNo [models.dev]
Structured outputYes [models.dev]
Hugging Facemistralai/Mistral-Small-24B-Instruct-2501 [Hugging Face]

Mistral Small 3 API pricing by provider

USD per million tokens. Sorted by input price.

ProviderInputOutputCached inputContextSource
DeepInfra
mistralai/Mistral-Small-24B-Instruct-2501
$0.05$0.08–33Kmodels.dev, 18 Nov 2026
OpenRouter
mistralai/mistral-small-24b-instruct-2501
$0.05$0.08–33KOpenRouter, 18 Nov 2026

Running Mistral Small 3 locally

Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.

QuantizationWeightsTotal, 8K contextTotal, 32K context
Q4_K_M (4-bit)14 GB16 GB20 GB
Q8_0 (8-bit)25 GB27 GB31 GB
FP16 / BF16 (16-bit)47 GB49 GB53 GB

At 4-bit with an 8K conversation, Mistral Small 3 needs about 16 GB, so it fits on a graphics card with 24 GB or more, such as the NVIDIA GeForce RTX 3090 24 GB. Check your own hardware, or see Best GPU for local LLMs on our sister site AIPCs.

Frequently asked questions

What is the context window of Mistral Small 3?

Mistral Small 3 has a context window of 32,768 tokens (33K) and can generate up to 16,384 tokens in a single response, according to models.dev.

How much does the Mistral Small 3 API cost?

Mistral Small 3 is listed from $0.05 per million input tokens and $0.08 per million output tokens (DeepInfra). Prices last verified 18 Nov 2026.

Is Mistral Small 3 open source?

Mistral Small 3 is an open-weight model: its weights are published on Hugging Face as mistralai/Mistral-Small-24B-Instruct-2501 under the apache-2.0 license. Check the license terms before commercial use.

When was Mistral Small 3 released?

Mistral AI released Mistral Small 3 on 30 Jan 2025, according to models.dev.

Which providers offer Mistral Small 3?

We track Mistral Small 3 on 2 providers: DeepInfra, OpenRouter.

How much VRAM does Mistral Small 3 need?

About 16 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 27 GB at 8-bit. The 4-bit model fits on a graphics card with 24 GB or more, such as the NVIDIA GeForce RTX 3090 24 GB. These are estimates from the model's published parameter count and architecture.

Similar models from other labs

Comparable on output type, weights, release date, price and context window. Prices are the lowest listed, USD per million tokens.

ModelLabReleasedContextInput $/1MOutput $/1MWeights
Qwen/Qwen2.5-7B-InstructAlibaba (Qwen)19 Sept 202433K$0.05$0.05Open
Phi 4Microsoft10 Jan 202516K$0.07$0.14Open
Qwen3 Reranker 8BAlibaba (Qwen)5 Jun 202533K$0.05–Open
Reka Flash 3Rekaai10 Mar 202566K$0.10$0.20Open
Gemma 3 12B ITGoogle DeepMind12 Mar 2025131K$0.05$0.10Open
Gemma 3 4B ITGoogle DeepMind12 Mar 2025131K$0.05$0.10Open

Change history

  • Now listed on DeepInfra

Sources