Written by Jakub Rusinowski · Last updated September 6, 2026
Model library → Ministral 3 → Ministral 3 14B
The largest Ministral 3 and the one with a hard reasoning number: the 14B reasoning variant scores 85% on AIME 2025. 9.3 GB at Q4_K_M fits a 12 GB card. Vision-capable, 128K context, Apache 2.0.
Ministral 3 14B needs about 9 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
| Parameters | 14 Billion |
| Context window | 131,072 |
| Architecture | Dense Transformer (vision) |
| Provider | Mistral AI |
| Licence | Apache 2.0 |
| Specified at | Q4_K_M |
| System RAM | 16 GB |
| Record updated | 2026-09-06 |
Apache-2.0 — commercial use permitted. Commercial use permitted. No usage restrictions beyond attribution.
Modelled on a reference NVIDIA RTX 4090 (24 GB), with no KV cache (this record has no published architecture). Speed figures are ESTIMATES from the memory-bandwidth roofline described on the methodology page, not benchmarks we ran — rows marked measured come from published or reader-submitted runs. VRAM here includes the KV cache, so it reads higher than the headline figure above, which does not.
| Quant | Weights | VRAM needed | Est. speed | Fit on 24 GB |
|---|---|---|---|---|
| Q2_K | 4.6 GB | 5.4 GB | ~108 tok/s (est.) | Fits comfortably |
| Q3_K_M | 6.0 GB | 6.8 GB | ~91 tok/s (est.) | Fits comfortably |
| Q4_K_M | 8.5 GB | 9.3 GB | ~70 tok/s (est.) | Fits comfortably |
| Q5_K_M | 9.9 GB | 10.7 GB | ~61 tok/s (est.) | Fits comfortably |
| Q6_K | 11.5 GB | 12.3 GB | ~55 tok/s (est.) | Fits comfortably |
| Q8_0 | 14.9 GB | 15.7 GB | ~44 tok/s (est.) | Fits comfortably |
| F16 | 28.0 GB | 28.8 GB | ~4 tok/s (est.) | Offloads to system RAM (slow) |
Want the memory numbers alone, at every quantization level and your own context length? Use the Ministral 3 14B VRAM calculator.
or compare on Vast.ai from $0.35/hr (typical low · varies)
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
The cheapest catalogued GPU that runs Ministral 3 14B is the Intel Arc B570 (10 GB).
Install Ollama, then run:
ollama run ministral-3:14bWeights on Hugging Face: mistralai/Ministral-3-14B-Instruct.
Quality scores as published by the model's authors or an independent evaluator — not throughput, and not measured by us.
| Benchmark | Score | Provenance |
|---|---|---|
| AIME 2025 | 85 / 100 % | vendor-claimed · https://mistral.ai/news/mistral-3/ |
Best for: reasoning, general purpose, multimodal, coding.
← All Ministral 3 models | VRAM calculator | Check your own hardware