Google
Gemma 4 12B It
Gemma 4 12B It is an open-weights 12B-parameter model by Google. It holds an Intelligence score of 54 on the local.ai leaderboard (2026-08 snapshot).
open weights
54
local.ai Intelligence
Parameters
12B
Benchmarks (local.ai snapshot)
HuggingFace metadata · updated 2026-07-20
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 7GQ6_K 10.8GQ8_0 13.2GFP16 24G| Quantization | ≈ File size | Note |
|---|---|---|
| IQ2_XXS | 3.2 GB | Extreme compression, largest quality loss |
| IQ3_XXS | 4.3 GB | Very small |
| Q3_K_M | 5.5 GB | Small |
| NVFP4 | 6 GB | NVIDIA 4-bit |
| Q4_K_S | 6.5 GB | Common |
| Q4_K_M | 7 GB | Most common sweet spot |
| Q5_K_M | 9.1 GB | Balanced |
| Q6_K | 10.8 GB | Near-lossless |
| Q8_0 | 13.2 GB | Near-lossless |
| FP16 | 24 GB | Original weights |
Actual files on HuggingFace
| File | Size |
|---|---|
| model.safetensors | 22.28 GB |
GGUF: unsloth/gemma-4-12b-it-GGUF · 576,401 downloads
| File | Size | |
|---|---|---|
| MTP/mtp-gemma-4-12b-it-BF16.gguf | 0.80 GB | download |
| MTP/mtp-gemma-4-12b-it-F16.gguf | 0.80 GB | download |
| MTP/mtp-gemma-4-12b-it-Q8_0.gguf | 0.43 GB | download |
| gemma-4-12b-it-BF16.gguf | 22.20 GB | download |
| gemma-4-12b-it-IQ4_NL.gguf | 6.26 GB | download |
| gemma-4-12b-it-IQ4_XS.gguf | 5.94 GB | download |
| gemma-4-12b-it-Q3_K_M.gguf | 5.30 GB | download |
| gemma-4-12b-it-Q3_K_S.gguf | 4.78 GB | download |
| gemma-4-12b-it-Q4_0.gguf | 6.28 GB | download |
| gemma-4-12b-it-Q4_1.gguf | 6.89 GB | download |
| gemma-4-12b-it-Q4_K_M.gguf | 6.63 GB | download |
| gemma-4-12b-it-Q4_K_S.gguf | 6.30 GB | download |
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.