Back to Models
MistralAIOpen Source

Ministral 3 8B (v25.12, Base/Instruct/Reasoning)

Ministral 3 8B is a multimodal large language model with reasoning functions developed by MistralAI. It supports a long context window of 256K and is released under the Apache 2.0 license.

Parameters

80.0B

Context Window

256K

License

Apache 2.0

Release Date

2025-12-02

API Pricing

API pricing for this model is not yet available

Strengths

  • Advanced multimodal support
  • 256K vast context
  • Open Apache 2.0 license

Weaknesses

  • Computational resource load due to 80B scale
  • Concerns about response speed due to inference specialization
  • Lack of verification data in Japanese

Use Cases

  • Complex logical reasoning tasks
  • Analysis of long documents
  • Processing of multimodal data

Deep Analysis

Architecture

Dense Transformer (8B)

Balanced Ministral 3 model

Context Window

128K tokens

API Pricing (Input/Output)

$0.10 / $0.10 per 1M tokens

Release Date

December 2, 2025

License

Open weights (v25.12)

Strengths

  • Extremely affordable pricing
  • 128K context window
  • Open weights
  • Good for edge and on-device deployment
  • Competitive with larger models on many tasks

Weaknesses

  • Smaller context than 14B variant
  • Less capable on complex reasoning
  • Limited multimodal support

Competitor Comparison

ModelArenaSWEGPQAPrice
Ministral 3 14B---$0.20/$0.20
Llama 3.2 8B---Free (open)
Gemma 2 9B---Free (open)

Ministral 3 8B is a balanced 8B parameter model in Mistral's Ministral 3 family, offering strong performance at extremely low cost ($0.10/1M tokens). Released December 2025 with 128K context.

Analysis generated: 2026-05-24