Vendor · Language models · Audio models

Mistral Models: All 41 Models Compared

Mistral AI has 41 models in the directory. The strongest is Mistral Large 4: Rank 34 to 43 of 95 in the overall index.

Models417 of them since January 2026

Best rank34Mistral Large 4, Rank 34 to 43 of 95

Cheapest API$0.10Ministral 3 3B, input per 1M tokens

Planfrom €7.13Mistral Vibe Pro Education, per month incl. VAT, free plan available

Most popular53Magistral Small, popularity index out of 100

Summary

Mistral develops open-weight models, APIs, and AI products. In September 2026, it raised €3B in a Series D round at a post-money valuation above €21B.

In the directory: 33 language models and 8 audio models. 33 of them with open weights that can run locally.

All 41 models from Mistral AI

Every model there ever was. Models with a profile lead to every benchmark, price and source.

41 models

Model type

Status

Weights

Profile

Language models · 33


Mistral Large 4	Oct 6, 2026	Preview	28Rank 34 to 43 of 95	$0.70 / $2.10	113	524K	19
Mistral Medium 3.5	Apr 28, 2026	Active	below 1Rank 78 to 91 of 95	$1.50 / $7.50	158	262K	37
Mistral Small 4	Mar 16, 2026	Active	below 1	$0.15 / $0.60	185	256K	39
Devstral 2	Dec 9, 2025	Active	n/a	n/a	n/a	262K	33
Devstral Small 2	Dec 9, 2025	Active	n/a	n/a	n/a	262K	39
Ministral 3 14B	Dec 2, 2025	Active	n/a	$0.20 / $0.20	81	262K	37
Ministral 3 3B	Dec 2, 2025	Active	n/a	$0.10 / $0.10	182	262K	36
Ministral 3 8B	Dec 2, 2025	Active	n/a	$0.15 / $0.15	92	262K	37
Mistral Large 3	Dec 2, 2025	Active	below 1Rank 87 to 94 of 95	$0.50 / $1.50	84	262K	32
Magistral Small 1.2	Sep 18, 2025	Active	n/a	n/a	98	131K	35
Voxtral Mini 3B 2507	Jul 15, 2025	Active	n/a	n/a	n/a	33K	39
Voxtral Small (latest)	Jul 15, 2025	Active	n/a	n/a	n/a	32K	16
Voxtral Small 24B	Jul 15, 2025	Active	n/a	$0.10 / $0.30	n/a	32K	24
Devstral Small	Jul 10, 2025	Active	n/a	n/a	n/a	128K	42
Mistral Small 3.2	Jun 20, 2025	Deprecated	n/a	n/a	57	131K	37
Magistral Small	Jun 10, 2025	Active	n/a	n/a	n/a	131K	53
Mistral Small 3.1	Mar 17, 2025	Deprecated	below 1	n/a	39	131K	18
Mistral Small 3.1 24B	Mar 17, 2025	Active	n/a	n/a	n/a	128K	16
Mistral Small 3	Jan 30, 2025	Deprecated	n/a	n/a	52	33K	35
Mistral Large 2.1	Nov 18, 2024	Active	n/a	n/a	n/a	131K	28
Pixtral Large (latest)	Nov 1, 2024	Active	n/a	n/a	n/a	128K	23
Ministral 8B	Oct 16, 2024	Deprecated	n/a	n/a	n/a	128K	17
Ministral 8B Instruct	Oct 16, 2024	Active	n/a	n/a	n/a	131K	19
Ministral 3Bno profile	October 2024	Active	n/a	n/a	n/a	n/a	n/a
Pixtral 12B	Sep 11, 2024	Active	n/a	n/a	n/a	128K	31
Mistral Large 2407	Jul 24, 2024	Active	n/a	n/a	n/a	131K	45
Mistral Nemo	Jul 18, 2024	Deprecated	n/a	n/a	n/a	131K	44
Codestral (latest)	May 29, 2024	Active	n/a	n/a	n/a	256K	16
Codestral-22B-v0.1	May 29, 2024	Active	n/a	n/a	n/a	33K	20
Mistral 7B v0.3	May 2024	Active	n/a	n/a	n/a	33K	40
Mixtral 8x22B	April 2024	Open source	n/a	$2 / $6	n/a	66K	33
Mixtral 8x7B	December 2023	Active	n/a	n/a	n/a	33K	40
Mistral 7Bno profile	September 2023	Active	n/a	n/a	n/a	n/a	n/a

Audio models · 8


Voxtral Mini Transcribe 3	Oct 20, 2026	Public preview	n/a	below 1
Voxtral Mini Transcribe Realtime 3	Oct 20, 2026	Public preview	n/a	below 1
Voxtral Mini TTS 3	Oct 20, 2026	Public preview	n/a	below 1
Voxtral Small Transcribe 3	Oct 20, 2026	Public preview	n/a	below 1
Voxtral Mini Transcribe 2	n/a	Active	n/a	26
Voxtral Mini Transcribe Realtime	n/a	Active	n/a	18
Voxtral Mini TTS	n/a	Active	n/a	29
Voxtral Mini TTS Realtime 3	n/a	Active	n/a	below 1

The current Mistral models explained

Mistral Medium 3.5

Language model · Apr 28, 2026

New in Mistral Medium 3.5 is the combination of 128 billion dense parameters, text and image input, and a 256,000-token context window.

What is distinctive about Mistral Small 4 is the combination of instruction following, reasoning, image understanding, and agentic coding. Its sparse mixture-of-experts design has 119 billion total parameters, 128 experts, and four active experts per token. It accepts text and images with up to 256,000 tokens of context; reasoning ranges from none to high.

Devstral 2

Language model · Dec 9, 2025

New in Devstral 2 is its focus on agents that operate development environments. The dense checkpoint has 123 billion parameters, 256,000 tokens of context, tool calls, repository navigation, and long code changes.

Devstral Small 2

Language model · Dec 9, 2025

What is distinctive about Devstral Small 2 is a 24B coding agent with text and image input. It handles 256,000 tokens of context, reads codebases, calls tools, and plans multi-step changes. The release uses Apache 2.0.

Ministral 3 14B

Language model · Dec 2, 2025

Ministral 3 14B is the largest compact variant in the open Mistral 3 family. Its dense text component has about 13.5 billion parameters plus a vision encoder. The model processes text and images with 256,000 tokens of context.

Ministral 3 3B

Language model · Dec 2, 2025

What is distinctive about Ministral 3 3B is that Mistral’s smallest open multimodal variant still brings 256,000 tokens of context and image and text input. Its text component has about 3.4 billion parameters plus a vision encoder.

Ministral 3 8B

Language model · Dec 2, 2025

New in Ministral 3 8B is the multimodal expansion of the compact edge class. Its text component has about 8.4 billion parameters plus a vision encoder; one local model handles image and text input with 256,000 tokens of context.

Mistral Large 3

Language model · Dec 2, 2025

New in Mistral Large 3 is Mistral’s first mixture-of-experts checkpoint since the Mixtral line. Mistral describes 675 billion total parameters and 41 billion active parameters with a 256,000-token context window. Text and images are inputs, and the weights use Apache 2.0.

Magistral Small 1.2

Language model · Sep 18, 2025

New in Magistral Small 1.2 is the multimodal revision of the 24B reasoning model. It enables longer reasoning. The weights use Apache 2.0.

Voxtral Mini 3B 2507

Language model · Jul 15, 2025

What is distinctive about Voxtral Mini 3B 2507 is audio understanding in a 3B checkpoint. It accepts audio and text with 32,000 tokens of context, transcribes recordings, answers questions, and creates summaries. It is the 3B variant alongside Voxtral Small 24B.

Voxtral Small (latest)

Language model · Jul 15, 2025

Voxtral Small (latest) is the rolling API entry for Mistral’s 24B audio-language model. It accepts audio and text and supports transcription, questions about recordings, and summaries. The alias corresponds to the fixed checkpoint Voxtral Small 24B.

Voxtral Small 24B

Language model · Jul 15, 2025

New in Voxtral Small 24B is the combination of audio transcription and content understanding. Mistral describes up to 30 minutes of transcription and up to 40 minutes of audio understanding per request, plus questions, summaries, and function calls across several languages. Its text backbone is based on Mistral Small 3.1.

17 more current models are in the model list above.

Which Mistral model fits?

Derived from the overall index, the API price and the measured speed of the current models.

The strongest choice

Mistral Large 4

The vendor's highest overall index

Overall index: 28 (Rank 34 to 43 of 95)
API per 1M tokens: $0.70 / $2.10
Speed: 113 tokens/s

The fastest

Mistral Small 4

185 tokens/s, a typical model manages 98

Overall index: below 1
API per 1M tokens: $0.15 / $0.60
Speed: 185 tokens/s

Mistral AI in the overall index

The 20 best language models, with the models from Mistral AI highlighted. “All” and the filter show every model in the index.

Overall index (Gradually Meta-Score v3)

20 of 95 models

Claude Opus 5.596.5

Gemini 4 Carbon96.1

Claude Fable 5.195.2

Claude Sonnet 5.593.9

Gemini 4 Argon93.5

GPT-6 Astra92.0

GPT-6.1 Sol90.0

Muse Wonder Full89.8

Claude Opus 587.8

Muse Wonder82.5

Kimi K3.1 Pro79.9

Claude Fable 579.8

GPT-6 Sol73.4

GPT-5.6 Sol71.3

Gemini 3.7 Flash69.6

Muse Spark 1.369.4

Gemini 3.8 Flash67.3

DeepSeek V4.1 Flash 101765.9

GPT-6.1 Terra64.2

Mistral Large 428.2

DeepSeek V4.1 Flash 1017

GPT-6.1 Terra

Mistral Large 4

Where does Mistral AI stand?

Models among the best 20 in the overall index

Anthropic5

OpenAI5

Google4

Meta3

DeepSeek1

Moonshot AI1

xAI1

Mistral AI0

Best Mistral AI model per area

Legal: Rank 9 to 40 of 67; Mistral Large 4
Finance: Rank 21 to 37 of 97; Mistral Large 4
Multimodal: Rank 21 to 26 of 47; Mistral Large 4
General: Rank 22 to 41 of 56; Mistral Large 4
Agents: Rank 26 to 39 of 82; Mistral Large 4
Reasoning: Rank 28 to 49 of 87; Mistral Large 4
Coding: Rank 29 to 46 of 89; Mistral Large 4
Health: Rank 33 to 46 of 69; Mistral Large 4
Mathematics: Rank 40 to 45 of 44; Mistral Large 2.1

Vendors compared

Vendor	Models	Model types	Best language model	Top 10	Open weights	New in 2026
Anthropic	33	33 language models	Claude Opus 5.5 (Rank 1 to 3)	4	0	13
OpenAI	63	46 language models, 4 image models, 2 video models, 9 audio models, 2 specialist models	GPT-6 Astra (Rank 4 to 8)	2	4	27
Google	66	43 language models, 4 image models, 6 video models, 11 audio models, 2 specialist models	Gemini 4 Carbon (Rank 1 to 5)	2	19	26
Meta	32	29 language models, 1 image model, 1 video model, 1 specialist model	Muse Wonder Full (Rank 6 to 9)	2	26	7
DeepSeek	23	23 language models	DeepSeek V4.1 Flash 1017 (Rank 13 to 20)	0	23	9
Moonshot AI	12	12 language models	Kimi K3.1 Pro (Rank 10 to 13)	0	10	7
xAI	16	12 language models, 2 image models, 2 video models	Grok 4.7 (Rank 18 to 24)	0	1	12
Mistral AI	41	33 language models, 8 audio models	Mistral Large 4 (Rank 34 to 43)	0	33	7

Strengths

Rank 34 to 43 of 95 in the overall index (Mistral Large 4)

Ministral 3 3B from $0.10 per 1M input tokens

33 models with open weights

Gaps

No image models or video models in the directory

5 models only with limited access or as a preview

Every Mistral AI model over time

By release date. Grey: retired models.

20232 models

SepMistral 7B

DecMixtral 8x7B

202412 models

AprMixtral 8x22B

MayMistral 7B v0.3

MayCodestral (latest)

MayCodestral-22B-v0.1

JulMistral Nemo

JulMistral Large 2407

SepPixtral 12B

OctMinistral 3B

OctMinistral 8B

OctMinistral 8B Instruct

NovPixtral Large (latest)

NovMistral Large 2.1

Releases in the last twelve months

6Dec 25

0Jan 26

0Feb 26

1Mar 26

1Apr 26

0May 26

0Jun 26

0Jul 26

0Aug 26

0Sep 26

5Oct 26

0Nov 26

7 releases since January 2026, by release date.

API prices

Prices per 1M tokens for the current language models, according to Mistral AI or the provider linked in each model profile.

Current model	Input	Output	Context
Mistral Large 4Preview	$0.70	$2.10	524K
Mistral Medium 3.5	$1.50	$7.50	262K
Mistral Large 3	$0.50	$1.50	262K
Mistral Small 4	$0.15	$0.60	256K
Ministral 3 3B	$0.10	$0.10	262K
Voxtral Small 24B	$0.10	$0.30	32K
Ministral 3 8B	$0.15	$0.15	262K
Ministral 3 14B	$0.20	$0.20	262K
Mixtral 8x22B	$2	$6	66K

Plans from Mistral AI

Monthly prices in Germany incl. VAT; “≈” is converted from US dollars. The vendor's models are often also part of other providers' plans, see the AI subscription comparison.

Plan	Per month	Note
Mistral Vibe Free	free
Mistral Vibe Pro Education	€7.13	with credits
Mistral Vibe Pro	€17.84	with credits
Mistral Vibe Team	€29.74	per seat, Team
Mistral Vibe Enterprise	n/a	Enterprise

Frequently asked questions about Mistral AI

What Mistral models are there?

The directory lists 41 Mistral models from Mistral AI: 33 language models and 8 audio models. It counts every model Mistral AI ever released, including retired ones; 39 of them have their own profile. 31 are currently available, the newest being Mistral Medium 3.5, Mistral Small 4 and Devstral 2.

Which Mistral model is the best?

Mistral Large 4. It is at rank 34 to 43 of 95 in the overall index, which combines independent benchmarks from several sources.

What is the latest Mistral model?

Voxtral Mini Transcribe 3, released on Oct 20, 2026. The timeline below shows every release by date.

How much do Mistral models cost?

Through the API per 1M tokens (input / output): Ministral 3 3B $0.10 / $0.10 and Mistral Large 4 $0.70 / $2.10. The prices of all 9 current language models are in the price table. As a subscription, Mistral starts at €7.13 per month incl. VAT (Mistral Vibe Pro Education).

Can I use Mistral for free?

Yes, with the Mistral Vibe Free plan. The cheapest paid plan is Mistral Vibe Pro Education at €7.13 per month incl. VAT.

Is Mistral better than ChatGPT?

In the overall index, no. The best OpenAI model, GPT-6 Astra (rank 4 to 8 of 95), is ahead of Mistral Large 4 (rank 34 to 43 of 95). Single tasks can differ.

Who makes Mistral?

Mistral is developed by Mistral AI. The company is based in Paris and was founded in 2023.

Can I run Mistral models locally?

Yes, 33 of the 41 models have open weights. Each model profile lists how much memory it needs.

Which Mistral model is the fastest?

Mistral Small 4 with 185 tokens per second according to Artificial Analysis. A typical model manages 98.

How large is the context window of the Mistral models?

The largest is Mistral Large 4 with 524,288 tokens (≈ 790 A4 pages). Every model's value is in the model list.

Sources and data date

Data as of Nov 1, 2026

Show all 10 sources

Mistral AI: mistral.ai/news/mistral-makes-sovereign-open-weight-ai-to-frontier/

Mistral Large 4: API prices

Mistral Medium 3.5: API prices

Ministral 3 14B: API prices

Ministral 3 3B: API prices

Ministral 3 8B: API prices

Voxtral Small 24B: API prices

Mixtral 8x22B: API prices

Artificial Analysis: output speed

AI subscription comparison (plans and sources)

All 41 models from Mistral AI

Language models · 33

Audio models · 8

The current Mistral models explained

Which Mistral model fits?

Mistral AI in the overall index

Overall index (Gradually Meta-Score v3)

Where does Mistral AI stand?

Models among the best 20 in the overall index

Best Mistral AI model per area

Vendors compared

Strengths

Gaps

Every Mistral AI model over time

20232 models

202412 models

202516 models

20267 models

No documented date4 models

Releases in the last twelve months

API prices

Plans from Mistral AI

Frequently asked questions about Mistral AI

Sources and data date