google · vision language

gemma-3-12b-it-qat-q4_0-unquantized

google/gemma-3-12b-it-qat-q4_0-unquantized

gemma-3-12b-it-qat-q4_0-unquantized at Q4_K_M is exactly 7,300,778,368 bytes (6.80 GiB / 7.30 GB) — an effective 4.792 bits per weight, not the nominal 4.

From the file· summed from 1 file(s)
Parameters
12.2B
Architecture
gemma3
Context
—
native (config.json)
License
gemma

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
UD-IQ1_S2.85 GiB3,057,668,6082.007—unsloth
UD-IQ1_M3.03 GiB3,249,207,8082.133—unsloth
UD-IQ2_XXS3.36 GiB3,604,238,8482.366—unsloth
IQ2_M4.01 GiB4,310,461,2162.829—bartowski
UD-IQ2_M4.07 GiB4,366,371,3282.866—unsloth
Q2_K_L4.44 GiB4,768,221,5683.130—unsloth
Q2_K4.44 GiB4,768,221,5683.130—unsloth
Q2_K4.44 GiB4,768,221,8563.130—bartowski
IQ3_XXS4.46 GiB4,784,900,8963.141—bartowski
UD-IQ3_XXS4.50 GiB4,834,206,2083.173—unsloth
Q2_K_L4.67 GiB5,012,075,2963.290—bartowski
IQ3_XS4.85 GiB5,206,166,1763.417—bartowski
Q3_K_S5.08 GiB5,458,315,6483.583—unsloth
Q3_K_S5.08 GiB5,458,315,9363.583—bartowski
IQ3_M5.27 GiB5,655,722,6563.712—bartowski
Q3_K_M5.60 GiB6,008,818,0483.944626unsloth
Q3_K_M5.60 GiB6,008,818,3363.944626bartowski
Q3_K_L6.04 GiB6,480,186,0164.254—bartowski
IQ4_XS6.10 GiB6,550,964,6084.300626unsloth
IQ4_XS6.10 GiB6,550,964,8964.300626bartowski
IQ4_NL6.41 GiB6,887,164,2884.521—unsloth
IQ4_NL6.41 GiB6,887,164,5764.521—bartowski
Q4_06.43 GiB6,909,282,6884.535626unsloth
Q4_06.43 GiB6,909,282,9764.535626bartowski
Q4_K_S6.46 GiB6,935,333,2484.553—unsloth
Q4_K_S6.46 GiB6,935,333,5364.553—bartowski
Q4_K_M6.80 GiB7,300,778,3684.792626unsloth
Q4_K_M6.80 GiB7,300,778,6564.792626bartowski
Q4_K_L7.03 GiB7,544,632,0964.952—bartowski
Q4_17.04 GiB7,559,563,6484.962—unsloth
Q4_17.04 GiB7,559,563,9364.962—bartowski
Q5_K_S7.67 GiB8,231,963,0085.404—unsloth
Q5_K_S7.67 GiB8,231,963,2965.404—bartowski
Q5_K_M7.87 GiB8,445,036,9285.543—unsloth
Q5_K_M7.87 GiB8,445,037,2165.543626bartowski
Q5_K_L8.09 GiB8,688,890,6565.704—bartowski
Q6_K9.00 GiB9,660,811,6486.341626unsloth
Q6_K9.00 GiB9,660,811,9366.341626bartowski
Q6_K_L9.22 GiB9,904,665,3766.502—bartowski
Q8_011.65 GiB12,510,212,8968.212626bartowski

Compare with

same modality, comparable size
vs Qwen3.5-9Bvs gemma-4-12B-itvs Qwen3.5-4Bvs Qwythos-9B-Claude-Mythos-5-1Mvs Qwen3.5-0.8Bvs gemma-4-E2B-itAll alternatives →Will it fit on my card? →

Will it run on your card?

full quant x context sweep
By VRAM:8GB12GB16GB24GB32GB48GB
Radeon RX 6400 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 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 6.38 GiB. The real file is 6.80 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

Architecture unavailable — this repository is gated and no ungated mirror was found. Exact file sizes above are still authoritative; only the KV math needs the config.

Questions people ask

How much VRAM does gemma-3-12b-it-qat-q4_0-unquantized need?
Q4_K_M is exactly 7,300,778,368 bytes (6.80 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of gemma-3-12b-it-qat-q4_0-unquantized 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.