Model family

Mistral Models

Mistral models cover multilingual workflows, coding, MoE architectures, vision-language experiments, and efficient local or hosted deployments.

Mistral AIUpdated 2026ChatCodeVisionMoELocal

Best for

Chat

Use this family hub to compare Mistral variants for chat workflows, then open the detail page for deeper deployment notes.

Code

Use this family hub to compare Mistral variants for code workflows, then open the detail page for deeper deployment notes.

Vision

Use this family hub to compare Mistral variants for vision workflows, then open the detail page for deeper deployment notes.

MoE

Use this family hub to compare Mistral variants for moe workflows, then open the detail page for deeper deployment notes.

Variants

Mistral models grouped by workflow

Coding

Reasoning

Vision / multimodal

Local-friendly

ChatPractical localchatefficient

Mistral Small 3.1

Mistral AI · Mistral

Best for: Builders who want a smaller capable model with permissive licensing and good local runtime support.

Local: A practical local candidate when quantized; still test memory use and context length on your hardware.

Details →
ChatOpen weights where releasedchatlocal

Mistral Large 2

Mistral AI · Mistral

Best for: Large multilingual assistant evaluation

Local: Use the exact checkpoint and quantization that matches your hardware and latency target.

Details →
ChatOpen weights where releasedchatlocal

Mistral 8x22B Instruct

Mistral AI · Mistral

Best for: MoE assistant evaluation

Local: Use the exact checkpoint and quantization that matches your hardware and latency target.

Details →
ChatOpen weights where releasedchatlocal

Mixtral 8x7B Instruct

Mistral AI · Mistral

Best for: MoE local and hosted baseline workflows

Local: Use the exact checkpoint and quantization that matches your hardware and latency target.

Details →
EdgeOpen weights where releasededgelocal

Mistral 7B Instruct

Mistral AI · Mistral

Best for: Local assistant baseline workflows

Local: Use the exact checkpoint and quantization that matches your hardware and latency target.

Details →
EdgeOpen weights where releasededgelocal

Ministral 8B

Mistral AI · Mistral

Best for: Efficient local workflows

Local: Use the exact checkpoint and quantization that matches your hardware and latency target.

Details →
EdgeOpen weights where releasededgelocal

Ministral 3B

Mistral AI · Mistral

Best for: Low-resource local workflows

Local: Use the exact checkpoint and quantization that matches your hardware and latency target.

Details →

Compare

All Mistral models in the directory

ModelTypeBest forLocal runner notesLicenseDetail
Mistral Small 3.1ChatBuilders who want a smaller capable model with permissive licensing and good local runtime support.A practical local candidate when quantized; still test memory use and context length on your hardware.Apache 2.0Open
Mistral Large 2ChatLarge multilingual assistant evaluationUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Pixtral LargeVisionVision-language assistant workflowsUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Pixtral 12BVisionVision-language local and hosted experimentsUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Codestral 22BCodeCoding assistant workflowsUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Mistral 8x22B InstructChatMoE assistant evaluationUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Mixtral 8x7B InstructChatMoE local and hosted baseline workflowsUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Mistral 7B InstructEdgeLocal assistant baseline workflowsUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Ministral 8BEdgeEfficient local workflowsUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Ministral 3BEdgeLow-resource local workflowsUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
Mathstral 7BReasoningMath and reasoning evaluationUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen
DevstralCodeDeveloper and coding workflow evaluationUse the exact checkpoint and quantization that matches your hardware and latency target.Check exact model cardOpen