Google

Gemma 4 E4B It

Gemma 4 E4B It is an open-weights 4B-parameter model by Google. It holds an Intelligence score of 28.4 on the local.ai leaderboard (2026-08 snapshot).

edgeopen weights
28.4
local.ai Intelligence
Parameters
4B
Benchmarks (local.ai snapshot)
τ²-bench
14%
GAIA
30%
GDPval
30%
HuggingFace metadata · updated 2026-07-20
repo
google/gemma-4-E4B-it
license
apache-2.0
downloads
5,321,979
likes
1,478

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 2.3GQ6_K 3.6GQ8_0 4.4GFP16 8G
Quantization≈ File sizeNote
IQ2_XXS1.1 GBExtreme compression, largest quality loss
IQ3_XXS1.4 GBVery small
Q3_K_M1.8 GBSmall
NVFP42 GBNVIDIA 4-bit
Q4_K_S2.2 GBCommon
Q4_K_M2.3 GBMost common sweet spot
Q5_K_M3 GBBalanced
Q6_K3.6 GBNear-lossless
Q8_04.4 GBNear-lossless
FP168 GBOriginal weights

Actual files on HuggingFace

FileSize
model.safetensors14.89 GB
GGUF: lmstudio-community/gemma-4-E4B-it-GGUF · 591,023 downloads
FileSize
gemma-4-E4B-it-Q4_K_M.gguf4.97 GBdownload
gemma-4-E4B-it-Q6_K.gguf5.79 GBdownload
gemma-4-E4B-it-Q8_0.gguf7.48 GBdownload
mmproj-gemma-4-E4B-it-BF16.gguf0.92 GBdownload
GGUF: unsloth/gemma-4-E4B-it-GGUF · 588,730 downloads
FileSize
MTP/mtp-gemma-4-E4B-it-BF16.gguf0.16 GBdownload
MTP/mtp-gemma-4-E4B-it-F16.gguf0.16 GBdownload
MTP/mtp-gemma-4-E4B-it-Q8_0.gguf0.09 GBdownload
gemma-4-E4B-it-BF16.gguf14.02 GBdownload
gemma-4-E4B-it-IQ4_NL.gguf4.50 GBdownload
gemma-4-E4B-it-IQ4_XS.gguf4.39 GBdownload
gemma-4-E4B-it-Q3_K_M.gguf3.78 GBdownload
gemma-4-E4B-it-Q3_K_S.gguf3.60 GBdownload
gemma-4-E4B-it-Q4_0.gguf4.50 GBdownload
gemma-4-E4B-it-Q4_1.gguf4.73 GBdownload
gemma-4-E4B-it-Q4_K_M.gguf4.64 GBdownload
gemma-4-E4B-it-Q4_K_S.gguf4.51 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.