Ministral 3 14B
Ministral 3 14B is Mistral AI's largest Ministral 3 model, offering base, instruct, and reasoning variants with vision support under Apache 2.0.
Model Overview
Capabilities, design details, and architectural traits
Ministral 3 14B - The Largest Model in the Ministral 3 Edge Family
Ministral 3 14B is the largest model in Mistral AI's Ministral 3 family, trained alongside the larger Mistral Large 3 as part of the same model generation. It is released as base, instruct, and reasoning variants, each with built-in image understanding, all under the Apache 2.0 license.
| Trait | Description |
|---|---|
| Three-variant release | Every Ministral 3 size, including the 14B, ships as base, instruct, and reasoning models, each with native vision support, openly licensed rather than closed |
| Token-efficient instruct variant | The instruct version is documented to match or exceed comparable models while often producing an order of magnitude fewer output tokens |
| Reasoning variant | A dedicated reasoning-trained version that thinks longer to raise accuracy on math, coding, and STEM-style tasks |
| Edge-first positioning | Designed primarily for local and on-device deployment rather than cloud-scale serving, while matching the performance of Mistral's larger Small 3.2 24B model |
| Generation sibling | Released together with Mistral Large 3, Mistral's mixture-of-experts flagship, as one coordinated model family |
Native Multimodal and Multilingual Design
Vision input and support for dozens of languages are built directly into the base architecture of every Ministral 3 14B variant, rather than added as a separate module or follow-up release. This lets the same checkpoint handle text, images, and cross-lingual tasks without switching models, which is the intended fit for enterprise and developer workflows that need one model for multiple input types at the edge.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Mistral | Anthropic | Anthropic |
| Release Date | December 2, 2025 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | - | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 262K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.20 Best Input Pricing | $4 | $2 |
| Output Pricing | $0.20 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | textimage | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 6.0 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 14.4 | - | - |
| Agentic Index | 1.1 | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from Mistral
Other models by Mistral
Mistral Large 4 Preview
Mistral Medium 3.5
Mistral Large 3
Mistral Medium 3.1
Top AI Models
Leading alternatives by intelligence score
