Google

Gemma 4 31B It

Gemma 4 31B It is an open-weights 31B-parameter model by Google. It holds an Intelligence score of 66.2 on the local.ai leaderboard (2026-08 snapshot).

open weights
66.2
local.ai Intelligence
Parameters
31B
Benchmarks (local.ai snapshot)
τ²-bench
77%
GAIA
71%
GDPval
61%
HuggingFace metadata · updated 2026-07-20
repo
google/gemma-4-31B-it
license
apache-2.0
downloads
9,965,914
likes
3,567

Metadata refreshed from the HuggingFace API (snapshot 2026-08-15T15:53:42.405Z). Context length and long-form descriptions are the next pipeline stage.

Does it fit your hardware?

Detect or select your GPU first. Green = weights + context headroom, yellow = tight, grey = does not fit.

Q4_K_M 18GQ6_K 27.9GQ8_0 34.1GFP16 62G
Quantization≈ File sizeNote
IQ2_XXS8.4 GBExtreme compression, largest quality loss
IQ3_XXS11.2 GBVery small
Q3_K_M14.3 GBSmall
NVFP415.5 GBNVIDIA 4-bit
Q4_K_S16.7 GBCommon
Q4_K_M18 GBMost common sweet spot
Q5_K_M23.6 GBBalanced
Q6_K27.9 GBNear-lossless
Q8_034.1 GBNear-lossless
FP1662 GBOriginal weights

Actual files on HuggingFace

FileSize
model-00001-of-00002.safetensors46.37 GB
model-00002-of-00002.safetensors11.89 GB
GGUF: unsloth/gemma-4-31B-it-GGUF · 549,239 downloads
FileSize
BF16/gemma-4-31B-it-BF16-00001-of-00002.gguf46.52 GBdownload
BF16/gemma-4-31B-it-BF16-00002-of-00002.gguf10.68 GBdownload
MTP/mtp-gemma-4-31B-it-BF16.gguf0.89 GBdownload
MTP/mtp-gemma-4-31B-it-F16.gguf0.89 GBdownload
MTP/mtp-gemma-4-31B-it-Q8_0.gguf0.48 GBdownload
gemma-4-31B-it-IQ4_NL.gguf16.10 GBdownload
gemma-4-31B-it-IQ4_XS.gguf15.25 GBdownload
gemma-4-31B-it-Q3_K_M.gguf13.72 GBdownload
gemma-4-31B-it-Q3_K_S.gguf12.30 GBdownload
gemma-4-31B-it-Q4_0.gguf16.15 GBdownload
gemma-4-31B-it-Q4_1.gguf17.81 GBdownload
gemma-4-31B-it-Q4_K_M.gguf17.07 GBdownload
GGUF: unsloth/gemma-4-31B-it-qat-GGUF · 312,475 downloads
FileSize
MTP/mtp-gemma-4-31B-it-BF16.gguf0.89 GBdownload
MTP/mtp-gemma-4-31B-it-F16.gguf0.89 GBdownload
MTP/mtp-gemma-4-31B-it-Q4_0.gguf0.26 GBdownload
MTP/mtp-gemma-4-31B-it-Q8_0.gguf0.48 GBdownload
gemma-4-31B-it-qat-UD-Q4_K_XL.gguf16.10 GBdownload
mmproj-BF16.gguf1.12 GBdownload
mmproj-F16.gguf1.12 GBdownload
mmproj-F32.gguf2.14 GBdownload
mtp-gemma-4-31B-it.gguf0.26 GBdownload

What you save self-hosting this model

Uses the model's own API price when it exists; otherwise the closest commercial equivalent.

Waiting for hardware detection…

RunLocal presents third-party leaderboard data (local.ai / Exo Labs, 2026-08) with its own fit and cost estimates. Verify pricing, licenses and file sizes at the source before deploying a model.