mlx-community/gemma-4-e2b-it-qat-OptiQ-4bit
Built with mlx-optiq, the MLX-native toolkit to quantize, fine-tune, and serve LLMs locally on Apple Silicon, no PyTorch and no cloud. Try the Lab · All OptiQ quants · Docs
A 4-bit mixed-precision MLX quant produced by mlx-optiq, built on Google's quantization-aware-trained (QAT) Gemma-4 base. OptiQ's sensitivity-guided per-layer bit allocation is applied on top of weights that were trained to survive low-bit quantization, and it still beats a uniform 4-bit quant of the same QAT base by +2.09 Capability Score points.
This is a quant of google/gemma-4-E2B-it-qat-q4_0-unquantized. Per-layer bit-widths come from a KL-divergence sensitivity pass on a six-domain calibration mix (prose, reasoning, code, agent, tool-call, constraint-bearing instructions). Sensitive layers go to 8-bit, robust ones stay at 4-bit.
Quantization details
| Property | Value |
|---|---|
| Base | google/gemma-4-E2B-it-qat-q4_0-unquantized (QAT) |
| Predominant precision | 4-bit |
| Components at 8-bit (sensitive) | 144 |
| Components at 4-bit (robust) | 132 |
| Total quantized components | 276 |
| Achieved bits-per-weight | 5.24 |
| Group size | 64 |
| Reference for sensitivity | bf16 |
| Calibration mix | six-domain mix |
| Vision | bf16 sidecar (optiq_vision.safetensors), image+text via optiq |
| Speculative drafter | google/gemma-4-E2B-it-qat-q4_0-unquantized-assistant via optiq serve --drafter |
Capability Score
Six-metric mean (MMLU, GSM8K, IFEval, BFCL, HumanEval, HashHop), scored against a uniform 4-bit quant of the same QAT base. That comparison isolates what the mixed-precision allocation adds, holding the base fixed.
| Benchmark | This model (OptiQ, QAT base) | Uniform-4 (QAT base) | Delta |
|---|---|---|---|
| MMLU (5-shot, 1000) | 48.5% | 46.7% | +1.8 |
| GSM8K (1000) | 58.6% | 56.2% | +2.4 |
| IFEval (full, strict) | 66.0% | 67.7% | -1.7 |
| BFCL-V3 simple (200) | 89.5% | 88.5% | +1.0 |
| HumanEval (pass@1, 164) | 62.8% | 59.8% | +3.0 |
| HashHop (long-context) | 18.0% | 12.0% | +6.0 |
| Capability Score (mean) | 57.23 | 55.14 | +2.09 |
OptiQ adds +2.09 points over uniform 4-bit on this QAT base, close to the +2.12 it adds on the vanilla gemma-4-e2b-it base, so the per-layer allocation keeps paying off even after QAT has made the weights more quantization-robust. The mixed quant is 5.24 bits-per-weight (about 4.9 GB on disk) versus 4.0 bits-per-weight (about 2.4 GB) for uniform 4-bit: the gain comes from spending the extra budget on the layers that need it.
Usage
mlx-lm loads it directly for text:
from mlx_lm import load, generate
model, tokenizer = load("mlx-community/gemma-4-e2b-it-qat-OptiQ-4bit")
print(generate(model, tokenizer, "Explain mixed-precision quantization.", max_tokens=256))
Image+text input and the speculative drafter run through mlx-optiq:
pip install mlx-optiq
optiq serve --model mlx-community/gemma-4-e2b-it-qat-OptiQ-4bit \
--drafter google/gemma-4-E2B-it-qat-q4_0-unquantized-assistant
The same repo loads text-only under stock mlx-lm and image+text under optiq. The bf16 vision tower rides in optiq_vision.safetensors, which mlx-lm ignores (it globs model*.safetensors), so both paths work from one artifact.
Quantize your own
This quant was produced by mlx-optiq. Point it at any Hugging Face model to get the same sensitivity-aware mixed precision:
pip install mlx-optiq
optiq convert <hf-model-id> --target-bpw 5.0 --candidate-bits 4,8
optiq lab # or open the full local workbench: chat, compare, quantize, fine-tune
License
Gemma Terms of Use. Built on google/gemma-4-E2B-it-qat-q4_0-unquantized.