mistralai · text

Ministral-3-14B-Instruct-2512

mistralai/Ministral-3-14B-Instruct-2512

Ministral-3-14B-Instruct-2512 at Q4_K_M is exactly 8,239,067,840 bytes (7.67 GiB / 8.24 GB) — an effective 4.727 bits per weight, not the nominal 4. Its KV cache at 32K is 5.00 GiB.

From the file· summed from 1 file(s)From the file· KV per layer
Parameters
13.9B
Architecture
mistral3
40 layers
Context
262,144
native (config.json)
License
apache-2.0

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
UD-IQ1_S3.21 GiB3,441,750,7201.974—unsloth
UD-IQ1_M3.42 GiB3,671,126,7202.106—unsloth
UD-IQ2_XXS3.78 GiB4,057,379,5202.328—unsloth
UD-IQ2_M4.57 GiB4,912,132,8002.818—unsloth
Q2_K4.89 GiB5,246,530,2403.010—unsloth
Q2_K_L5.03 GiB5,403,816,6403.100—unsloth
UD-IQ3_XXS5.12 GiB5,493,437,1203.151—unsloth
Q3_K_S5.66 GiB6,074,905,2803.485—unsloth
Q3_K_M6.22 GiB6,682,096,3203.833—unsloth
IQ4_XS6.92 GiB7,432,155,8404.264—unsloth
NVFP47.25 GiB7,779,496,8004.463—FreedomAISVR
Q4_07.27 GiB7,805,711,0404.478—unsloth
IQ4_NL7.27 GiB7,805,711,0404.478—unsloth
Q4_K_S7.30 GiB7,834,546,8804.495—unsloth
Q4_K_M7.67 GiB8,239,067,8404.727—unsloth
Q4_K_M7.67 GiB8,239,593,0244.727—mistralai
Q4_17.99 GiB8,581,657,2804.923—unsloth
Q5_K_S8.74 GiB9,383,817,9205.383—unsloth
Q5_K_M8.96 GiB9,620,566,7205.519—unsloth
Q5_K_M8.96 GiB9,621,091,9045.519—mistralai
Q6_K10.33 GiB11,088,409,2806.361—unsloth
Q8_013.37 GiB14,359,311,0408.238—unsloth
Q8_013.37 GiB14,359,836,2248.238—mistralai
BF1625.17 GiB27,020,865,95215.501—unsloth
BF1625.17 GiB27,021,391,42415.502—mistralai

KV cache by context

computed per layer
ContextKV cache (f16)●Flat formulaOverstated byFull / windowed / recurrent
4,0960.63 GiB0.63 GiB—40 / 0 / 0
8,1921.25 GiB1.25 GiB—40 / 0 / 0
16,3842.50 GiB2.50 GiB—40 / 0 / 0
32,7685.00 GiB5.00 GiB—40 / 0 / 0
65,53610.00 GiB10.00 GiB—40 / 0 / 0
131,07220.00 GiB20.00 GiB—40 / 0 / 0

Compare with

same modality, comparable size
vs Qwen3-Coder-30B-A3B-Instructvs gemma-4-12B-it-qat-q4_0-unquantizedvs gemma-4-E4B-itvs Qwen3-30B-A3B-Thinking-2507vs Qwen3-4Bvs Qwen3-8BAll alternatives →Will it fit on my card? →

Will it run on your card?

full quant x context sweep
By VRAM:8GB12GB16GB24GB32GB48GB
Radeon RX 6500 XT 4GBGeForce RTX 3050 6GBGeForce RTX 5050 8GBGeForce RTX 3080 10GBGeForce RTX 2080 Ti 11GBGeForce RTX 5070 12GBGeForce RTX 5060 Ti 16GBApple M3 Pro 18GBGeForce RTX 3080 Ti 20GBGeForce RTX 5090 D V2 24GBGeForce RTX 5090 D 32GBApple M5 Max 36GBApple M5 Pro 48GBApple M5 Max 64GBApple M3 Ultra 96GBApple M5 Max 128GBApple M2 Ultra 192GBApple M3 Ultra 256GBApple M3 Ultra 512GB

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 7.31 GiB. The real file is 7.67 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json
Layers
40
Attention heads
32
KV heads
8
Head dim
128
Hidden size
5120
Vocab
131,072
Sliding window
none
SWA period
—
MLA
no
Experts
—
Experts per token
—
use_sliding_window
—

Questions people ask

How much VRAM does Ministral-3-14B-Instruct-2512 need?
Q4_K_M is exactly 8,239,067,840 bytes (7.67 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is Ministral-3-14B-Instruct-2512's KV cache?
5.00 GiB at 32K context with an f16 cache, computed per layer. Quantizing the cache to q8_0 roughly halves it, which is often the difference between a context length fitting and not.
Which quantization of Ministral-3-14B-Instruct-2512 should I use?
Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.