DEVELOPER REFERENCE · VERIFIED MODEL EVIDENCE

Compare 47 source-verified model records, repository licenses, published weight formats and dated evidence. No overall model ranking.

Evidence statusSource-first verified
Latest record check2026-11-13
Latest record full verification2026-11-03

DIRECT ANSWER · SOURCE-OBSERVED REGISTRY

What Mistral AI model evidence is available?

Open Model Weights currently tracks 47 field-verified Mistral AI model records. In this indexed sample, the recorded modalities are 26 Text; 4 Not declared; 4 Time-series forecasting; 3 Speculative decoding asset; 2 Image + text → text; 2 Text / code; 1 Automatic speech recognition; 1 Safety / text; 1 Text + image → text; 1 Text / reasoning; 1 Text and images → text; 1 Text-to-speech. The published repository license labels include 37 apache-2.0; 4 other; 4 Timestral Modified MIT (model weights); 1 cc-by-nc-4.0; 1 Mistral Ecosystem Open-Weight License v26.10. The listed weight formats include Safetensors (41), GGUF (5), PyTorch (4), ONNX (1).

Coverage describes the model repositories indexed here, not every model released by the publisher. A checked source, an unknown field and an unchanged source revision are different states. No missing value is treated as a negative answer.

47verified model records
12recorded modalities
5distinct license labels
47model repository links

DEVELOPER HQ / PRIMARY ORGANIZATION LOCATION

Primary organization location

Paris, France

Mistral AI

This location describes the developer or primary organization. It is not the nationality of a model, dataset, contributor, training run or infrastructure.

SOURCE-BACKED PUBLISHER PROFILE

Mistral AI: models, research and access

Mistral AI is a model developer and AI platform provider. Its official Mistral 3 launch describes downloadable Apache-2.0 models spanning compact Ministral variants and the large Mixture-of-Experts Mistral Large 3 family.

The index also includes older Mistral/Mixtral models, Devstral code agents, Pixtral vision, Voxtral speech and specialized speculative-decoding artifacts. Access and model licenses differ between these families and API-only services.

01

Mistral Large versus Ministral

Mistral's Mistral 3 launch describes Ministral 3 sizes of 3B, 8B and 14B, plus a large Mixture-of-Experts model. Inference hardware and costs depend on artifacts and precision. The registry includes multiple base, instruct, reasoning and quantized variants, which should be assessed as distinct checkpoints.

02

Code, vision and speech specialization

Devstral focuses on software-development workflows; Pixtral records process images; Voxtral includes audio recognition and text-to-speech. These are task-specific models, not directly interchangeable with general-purpose chat checkpoints. Speculative-decoding EAGLE assets are supporting artifacts, not independent language-model alternatives.

03

Licensing across releases

Mistral states that the Mistral 3 family launched under Apache 2.0. The current evidence registry also includes custom-license entries and the non-commercial CC BY-NC 4.0 label for Voxtral-4B-TTS. A permissive license on one release is not a blanket permission for all Mistral model weights or hosted features.

04

Private infrastructure and API are different scopes

Mistral offers hosted services while also publishing self-deployable weights for certain checkpoints. Its sovereignty announcement distinguishes regional inference endpoints and compute control. Neither French organizational origin nor a downloadable model establishes a user's data-residency or GDPR outcome; these depend on the exact setup and contracts.

Publisher background is attributed to official sources. Model counts and license classifications come from OMW repository evidence; no model benchmark or processing location is inferred.

MODEL-BY-MODEL REFERENCE

Which Mistral AI model is intended for your task?

Browse by published modality and licensing evidence. These are navigation cues, not quality, speed or suitability rankings.

Text + image → textapache-2.0

Mistral Small 4 119B A6B

General multimodal language

Check Mistral Small 4 for text-plus-image tasks and the real serving requirements of its MoE configuration.

Commercial use: allowedLicense evidence →
Text-to-speechcc-by-nc-4.0

Voxtral-4B-TTS-2603

Text-to-speech with non-commercial label

Check the CC BY-NC 4.0 license carefully; it differs from the Apache-2.0 label on other Mistral entries.

Commercial use: not-allowedLicense evidence →

RELATED RESEARCH & RELEASES

Source-backed publisher coverage

News about additional products is not automatically a field-verified registry addition.

FIELD-BY-FIELD COMPARISON · PRIMARY LINKS

Model evidence at a glance.

Compare the exact entries in our registry. Parameter values are taken from the source-checked record; a license label is not a legal opinion, and recognized weight files are not an inference-speed benchmark. Open each record for the underlying citations and observed history.

Model and parameter countModalityLicense evidenceFormats and filesSource and history
Mistral-7B-Instruct-v0.37.24802B published parametersNot declaredapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors5 filesRepository ↗Model evidence →History →
Voxtral-Mini-4B-Realtime-26024B published parametersAutomatic speech recognitionapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors2 filesRepository ↗Model evidence →History →
Mistral-7B-Instruct-v0.27.24173B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →PyTorch, Safetensors8 filesRepository ↗Model evidence →History →
Mistral-7B-v0.17.24173B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →PyTorch, Safetensors6 filesRepository ↗Model evidence →History →
Ministral-3-14B-Reasoning-251214B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors8 filesRepository ↗Model evidence →History →
Mixtral-8x7B-Instruct-v0.146.7028B published parametersNot declaredapache-2.0allowed commercial-use classificationInspect license evidence →PyTorch, Safetensors28 filesRepository ↗Model evidence →History →
Ministral-3-14B-Instruct-251214B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors6 filesRepository ↗Model evidence →History →
Mistral-Small-3.1-24B-Instruct-250324.0114B published parametersNot declaredapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors12 filesRepository ↗Model evidence →History →
Mistral-Small-3.2-24B-Instruct-250624B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors12 filesRepository ↗Model evidence →History →
Mistral-Nemo-Instruct-240712.2478B published parametersNot declaredapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors7 filesRepository ↗Model evidence →History →
Ministral-3-8B-Instruct-25129B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors5 filesRepository ↗Model evidence →History →
Ministral-3-8B-Instruct-2512-GGUF8B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →GGUF5 filesRepository ↗Model evidence →History →
Shieldstral-1.0-3B4B published parametersSafety / textapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors2 filesRepository ↗Model evidence →History →
Ministral-3-3B-Instruct-25124B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors2 filesRepository ↗Model evidence →History →
Ministral-3-3B-Instruct-2512-GGUF3B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →GGUF5 filesRepository ↗Model evidence →History →
Ministral-3-3B-Instruct-2512-BF164B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors4 filesRepository ↗Model evidence →History →
Ministral-3-14B-Instruct-2512-BF1614B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors8 filesRepository ↗Model evidence →History →
Ministral-3-14B-Instruct-2512-GGUF14B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →GGUF5 filesRepository ↗Model evidence →History →
Ministral-3-8B-Instruct-2512-BF169B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors6 filesRepository ↗Model evidence →History →
Mistral Small 4 119B A6B119B total / 6.5B active published parametersText + image → textapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors12 filesRepository ↗Model evidence →History →
Magistral-Small-250924B published parametersText / reasoningapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors12 filesRepository ↗Model evidence →History →
Timestral-209M0.209312B published parametersTime-series forecastingTimestral Modified MIT (model weights)allowed-with-conditions commercial-use classificationLicense file ↗Safetensors1 filesRepository ↗Model evidence →History →
Devstral-Small-2-24B-Instruct-251224B published parametersText / codeapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors10 filesRepository ↗Model evidence →History →
Timestral-28M0.0278212B published parametersTime-series forecastingTimestral Modified MIT (model weights)allowed-with-conditions commercial-use classificationLicense file ↗Safetensors1 filesRepository ↗Model evidence →History →
Ministral-3-3B-Reasoning-25124B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors4 filesRepository ↗Model evidence →History →
Mistral-Medium-3.5-128B128B published parametersTextotherunknown commercial-use classificationLicense file ↗Safetensors8 filesRepository ↗Model evidence →History →
Ministral-3-8B-Reasoning-25129B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors6 filesRepository ↗Model evidence →History →
Timestral-5M0.00520896B published parametersTime-series forecastingTimestral Modified MIT (model weights)allowed-with-conditions commercial-use classificationLicense file ↗Safetensors1 filesRepository ↗Model evidence →History →
Ministral-3-14B-Base-251214B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors8 filesRepository ↗Model evidence →History →
Mistral Large 4 (1T-A52B)1050B total / 52B active (publisher-declared) published parametersText and images → textMistral Ecosystem Open-Weight License v26.10allowed-with-conditions commercial-use classificationLicense file ↗Safetensors55 filesRepository ↗Model evidence →History →
Ministral-3-3B-Base-25124B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors4 filesRepository ↗Model evidence →History →
Timestral-123M0.122607B published parametersTime-series forecastingTimestral Modified MIT (model weights)allowed-with-conditions commercial-use classificationLicense file ↗Safetensors1 filesRepository ↗Model evidence →History →
Ministral-3-8B-Base-25129B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors6 filesRepository ↗Model evidence →History →
Ministral-3-8B-Reasoning-2512-GGUF8B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →GGUF5 filesRepository ↗Model evidence →History →
Pixtral-12B-240912B published parametersImage + text → textapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors1 filesRepository ↗Model evidence →History →
Devstral-2-123B-Instruct-2512125B published parametersText / codeotherunknown commercial-use classificationLicense file ↗Safetensors56 filesRepository ↗Model evidence →History →
Mistral-Large-3-675B-Instruct-2512-NVFP4675B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors274 filesRepository ↗Model evidence →History →
Ministral-3-3B-Reasoning-2512-GGUF3B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →GGUF5 filesRepository ↗Model evidence →History →
Mistral-Large-3-675B-Instruct-2512675B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors273 filesRepository ↗Model evidence →History →
Voxtral-4B-TTS-26034B published parametersText-to-speechcc-by-nc-4.0not-allowed commercial-use classificationInspect license evidence →PyTorch, Safetensors21 filesRepository ↗Model evidence →History →
Ministral-3-3B-Instruct-2512-ONNX4B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →ONNX15 filesRepository ↗Model evidence →History →
Mistral-Small-4-119B-2603-eagle119B published parametersSpeculative decoding assetapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors1 filesRepository ↗Model evidence →History →
Leanstral-1.5-119B-A6B119B published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors8 filesRepository ↗Model evidence →History →
Mistral-Medium-3.5-128B-EAGLE128B published parametersSpeculative decoding assetotherunknown commercial-use classificationLicense file ↗Safetensors1 filesRepository ↗Model evidence →History →
Mistral-Large-3-675B-Instruct-2512-Eagle675B published parametersSpeculative decoding assetapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors1 filesRepository ↗Model evidence →History →
Pixtral-Large-Instruct-2411— published parametersImage + text → textotherunknown commercial-use classificationInspect license evidence →Safetensors53 filesRepository ↗Model evidence →History →
Leanstral-2603— published parametersTextapache-2.0allowed commercial-use classificationInspect license evidence →Safetensors8 filesRepository ↗Model evidence →History →

LICENSE & COVERAGE · NOT LEGAL ADVICE

How do the recorded licenses differ?

Different models from the same developer can use different license terms. The distribution below describes the labels and field-level classifications observed in the listed repositories; it does not automatically establish permissions for a specific deployment.

License labels

  • apache-2.037 of 47
  • other4 of 47
  • Timestral Modified MIT (model weights)4 of 47
  • cc-by-nc-4.01 of 47
  • Mistral Ecosystem Open-Weight License v26.101 of 47

Commercial-use classifications

  • allowed37 of 47
  • allowed-with-conditions5 of 47
  • unknown4 of 47
  • not-allowed1 of 47

Model modalities

  • Text26 of 47
  • Not declared4 of 47
  • Time-series forecasting4 of 47
  • Speculative decoding asset3 of 47
  • Image + text → text2 of 47
  • Text / code2 of 47
  • Automatic speech recognition1 of 47
  • Safety / text1 of 47
  • Text + image → text1 of 47
  • Text / reasoning1 of 47
  • Text and images → text1 of 47
  • Text-to-speech1 of 47

A classification of unknown is not a prohibition. Even a recognized permissive license can carry conditions. Review the license for the exact model and revision. Open the license explorer →

EVIDENCE METHOD · TRANSPARENT LIMITS

How this developer profile is assembled.

We aggregate the current source-checked model records, rather than publish an unverified corporate biography. Repository metadata, model cards, license files and recognized weight artifacts supply the facts where available. The record-check date does not assert that a model or its license changed on that date.

Source-derived runtime signals are not hardware benchmarks. Weight-only memory estimates omit runtime overhead, activations and other deployment requirements. Model-specific uncertainties remain visible in the individual evidence records.

QUESTIONS · DIRECT ANSWERS

Frequently asked questions about Mistral AI models.

Which Mistral AI models are covered?

Open Model Weights currently lists 47 field-verified records for Mistral AI. Their recorded model modalities are: 26 Text; 4 Not declared; 4 Time-series forecasting; 3 Speculative decoding asset; 2 Image + text → text; 2 Text / code; 1 Automatic speech recognition; 1 Safety / text; 1 Text + image → text; 1 Text / reasoning; 1 Text and images → text; 1 Text-to-speech. This is a registry sample, not an exhaustive publisher catalog.

Which licenses are recorded for Mistral AI models?

In the 47 indexed records, repository license labels are: 37 apache-2.0; 4 other; 4 Timestral Modified MIT (model weights); 1 cc-by-nc-4.0; 1 Mistral Ecosystem Open-Weight License v26.10. License labels do not replace the model-specific terms or establish rights for every use case.

Where can I verify Mistral AI model information?

Each indexed model links to its official source repository and a field-level Open Model Weights profile. The most recent record checks shown here reach 2026-11-13; checking a repository does not imply that its contents changed.

Which Mistral AI models have Apache-2.0 weights?

Mistral's official Mistral 3 announcement identifies Apache 2.0 for Mistral Large 3 and the 3B/8B/14B Ministral family. Other families can have different terms; check each record.

What is the distinction between Devstral and Voxtral?

The listed Devstral checkpoints target coding workflows, whereas Voxtral records cover speech recognition or synthesis. Different model tasks require different evaluations and runtime integrations.

Do Mistral AI's European operations guarantee EU-only processing?

No. Company location, a model's open-weight status and a hosted endpoint's actual data region are separate facts. Verify the deployment, service agreement and region when data residency is required.

COMPLETE INDEXED SELECTION

All 47 indexed model profiles.

Mistral-7B-Instruct-v0.3
Not declared
32,768 tokens
apache-2.0
Field-verified
Voxtral-Mini-4B-Realtime-2602
Automatic speech recognition
131,072 tokens
apache-2.0
Field-verified
Mistral-7B-Instruct-v0.2
Text
32,768 tokens
apache-2.0
Field-verified
Mistral-7B-v0.1
Text
32,768 tokens
apache-2.0
Field-verified
Ministral-3-14B-Reasoning-2512
Text
262,144 tokens
apache-2.0
Field-verified
Mixtral-8x7B-Instruct-v0.1
Not declared
32,768 tokens
apache-2.0
Field-verified
Ministral-3-14B-Instruct-2512
Text
262,144 tokens
apache-2.0
Field-verified
Mistral-Small-3.1-24B-Instruct-2503
Not declared
131,072 tokens
apache-2.0
Field-verified
Mistral-Small-3.2-24B-Instruct-2506
Text
131,072 tokens
apache-2.0
Field-verified
Mistral-Nemo-Instruct-2407
Not declared
131,072 tokens
apache-2.0
Field-verified
Ministral-3-8B-Instruct-2512
Text
262,144 tokens
apache-2.0
Field-verified
Ministral-3-8B-Instruct-2512-GGUF
Text
256,000 tokens
apache-2.0
Field-verified
Shieldstral-1.0-3B
Safety / text
262,144 tokens
apache-2.0
Field-verified
Ministral-3-3B-Instruct-2512
Text
262,144 tokens
apache-2.0
Field-verified
Ministral-3-3B-Instruct-2512-GGUF
Text
256,000 tokens
apache-2.0
Field-verified
Ministral-3-3B-Instruct-2512-BF16
Text
262,144 tokens
apache-2.0
Field-verified
Ministral-3-14B-Instruct-2512-BF16
Text
262,144 tokens
apache-2.0
Field-verified
Ministral-3-14B-Instruct-2512-GGUF
Text
256,000 tokens
apache-2.0
Field-verified
Ministral-3-8B-Instruct-2512-BF16
Text
262,144 tokens
apache-2.0
Field-verified
Mistral Small 4 119B A6B
Text + image → text
1,048,576 tokens
apache-2.0
Field-verified
Magistral-Small-2509
Text / reasoning
131,072 tokens
apache-2.0
Field-verified
Timestral-209M
Time-series forecasting
32,768 historical time steps (not language tokens)
Timestral Modified MIT (model weights)
Field-verified
Devstral-Small-2-24B-Instruct-2512
Text / code
393,216 tokens
apache-2.0
Field-verified
Timestral-28M
Time-series forecasting
32,768 historical time steps (not language tokens)
Timestral Modified MIT (model weights)
Field-verified
Ministral-3-3B-Reasoning-2512
Text
262,144 tokens
apache-2.0
Field-verified
Mistral-Medium-3.5-128B
Text
262,144 tokens
other
Field-verified
Ministral-3-8B-Reasoning-2512
Text
262,144 tokens
apache-2.0
Field-verified
Timestral-5M
Time-series forecasting
32,768 historical time steps (not language tokens)
Timestral Modified MIT (model weights)
Field-verified
Ministral-3-14B-Base-2512
Text
262,144 tokens
apache-2.0
Field-verified
Mistral Large 4 (1T-A52B)
Text and images → text
1,048,576 tokens
Mistral Ecosystem Open-Weight License v26.10
Field-verified
Ministral-3-3B-Base-2512
Text
262,144 tokens
apache-2.0
Field-verified
Timestral-123M
Time-series forecasting
32,768 historical time steps (not language tokens)
Timestral Modified MIT (model weights)
Field-verified
Ministral-3-8B-Base-2512
Text
262,144 tokens
apache-2.0
Field-verified
Ministral-3-8B-Reasoning-2512-GGUF
Text
256,000 tokens
apache-2.0
Field-verified
Pixtral-12B-2409
Image + text → text
Not declared in checked config/model card
apache-2.0
Field-verified
Devstral-2-123B-Instruct-2512
Text / code
262,144 tokens
other
Field-verified
Mistral-Large-3-675B-Instruct-2512-NVFP4
Text
256,000 tokens
apache-2.0
Field-verified
Ministral-3-3B-Reasoning-2512-GGUF
Text
256,000 tokens
apache-2.0
Field-verified
Mistral-Large-3-675B-Instruct-2512
Text
256,000 tokens
apache-2.0
Field-verified
Voxtral-4B-TTS-2603
Text-to-speech
Not declared in checked config/model card
cc-by-nc-4.0
Field-verified
Ministral-3-3B-Instruct-2512-ONNX
Text
262,144 tokens
apache-2.0
Field-verified
Mistral-Small-4-119B-2603-eagle
Speculative decoding asset
256,000 tokens
apache-2.0
Field-verified
Leanstral-1.5-119B-A6B
Text
256,000 tokens
apache-2.0
Field-verified
Mistral-Medium-3.5-128B-EAGLE
Speculative decoding asset
256,000 tokens
other
Field-verified
Mistral-Large-3-675B-Instruct-2512-Eagle
Speculative decoding asset
256,000 tokens
apache-2.0
Field-verified
Pixtral-Large-Instruct-2411
Image + text → text
128,000 tokens
other
Field-verified
Leanstral-2603
Text
256,000 tokens
apache-2.0
Field-verified