Gemma 4 E2B It RotorQuant MLX 2bit

[!TIP] KV-cache quantization without any fork (recommended, 2026): upstream llama.cpp/Ollama now cover this natively — use -ctk q8_0 -ctv q8_0 (~half KV memory, negligible quality loss: perplexity +0.002–0.05) or -ctk q4_0 -ctv q4_0 (~quarter memory, ≈7.6% perplexity increase). In Ollama: OLLAMA_KV_CACHE_TYPE=q8_0 with OLLAMA_FLASH_ATTENTION=1. Keep K and V types symmetric to stay on the fast fused Flash-Attention path. Since April 2026, mainline llama.cpp also applies Hadamard rotation to KV activations (PR #21038), which greatly improves low-bit KV quality (opt-out: LLAMA_ATTN_ROT_DISABLE=1).

The RotorQuant/TurboQuant fork flow below is experimental/legacy: the TurboQuant llama.cpp PR was closed without merging (June 2026) and the fork is unmaintained relative to mainline. It is NOT required to use this model.

Gemma 4 E2B-it - RotorQuant MLX 2-bit

2-bit weight-quantized MLX version of google/gemma-4-E2B-it with RotorQuant KV-cache quantization. Optimized for Apple Silicon inference via the MLX framework. RotorQuant delivers 5.3x faster prefill and 28% faster decode compared to TurboQuant. The most aggressive quantization, fitting the full model in the smallest possible footprint.

Approximate model size: ~0.6 GB

Model Specifications

Property	Value
Base Model	google/gemma-4-E2B-it
Parameters	~2 billion
Architecture	Dense transformer
Modality	Multimodal: image + text input, text output
License	Apache 2.0
Weight Quantization	2-bit (~0.6 GB)
KV-Cache Quantization	RotorQuant
Framework	MLX (Apple Silicon)

Quickstart

import mlx.core as mx
from mlx_lm import load, generate

model, tokenizer = load("majentik/gemma-4-E2B-it-RotorQuant-MLX-2bit")

prompt = "Describe this image in detail."
response = generate(model, tokenizer, prompt=prompt, max_tokens=512)
print(response)

For multimodal usage with images:

from mlx_vlm import load, generate

model, processor = load("majentik/gemma-4-E2B-it-RotorQuant-MLX-2bit")

prompt = "What do you see in this image?"
output = generate(model, processor, prompt=prompt, image="path/to/image.jpg", max_tokens=512)
print(output)

What is RotorQuant?

RotorQuant is a high-performance KV-cache quantization method that achieves significantly better throughput than TurboQuant. Combined with 2-bit weight quantization in MLX, this provides maximum compression with the best available KV-cache performance: the smallest possible model footprint plus the fastest compressed KV cache for efficient long-context generation.

Key advantages over TurboQuant:

• 5.3x faster prefill
• 28% faster decode
• Equivalent memory savings

Note: 2-bit quantization is the most aggressive option and may result in some quality degradation compared to higher-precision variants. It is best suited for experimentation, rapid prototyping, or hardware-constrained environments.

KV-Cache Quantization Comparison

Method	Prefill Speed	Decode Speed	Memory Savings	Reference
TurboQuant	1x (baseline)	1x (baseline)	High	arXiv: 2504.19874
RotorQuant	5.3x faster	28% faster	High	GitHub

Memory Estimates (Gemma 4 E2B-it)

Precision	Approximate Size	MLX Variant
FP16 (original)	~4 GB	--
8-bit quantized	~2 GB	RotorQuant-MLX-8bit
4-bit quantized	~1.2 GB	RotorQuant-MLX-4bit
2-bit quantized	~0.6 GB	This model

Hardware Requirements

This model requires approximately 0.6 GB of unified memory. Recommended hardware:

• Apple M1 (8 GB+)
• Apple M2 (8 GB+)
• Apple M3 (8 GB+)
• Apple M4 (8 GB+)
• Any Apple Silicon Mac with 8 GB+ unified memory

See Also

• google/gemma-4-E2B-it -- Base model
• majentik/gemma-4-E2B-it-RotorQuant -- RotorQuant KV-cache only (transformers)
• majentik/gemma-4-E2B-it-RotorQuant-MLX-8bit -- MLX 8-bit variant
• majentik/gemma-4-E2B-it-RotorQuant-MLX-4bit -- MLX 4-bit variant
• majentik/gemma-4-E2B-it-TurboQuant-MLX-2bit -- TurboQuant MLX 2-bit variant
• RotorQuant GitHub
• MLX Framework

Quant trade-off (MLX lane)

Bits	Approx size	Use case	Recommendation
2-bit	~532 MB	Aggressive quantization	Very low-RAM Macs
3-bit	~737 MB	Lossy but small	Low-RAM Macs
4-bit	~860 MB	Balanced default	Recommended for most Macs
5-bit	~1.0 GB	Higher fidelity	Quality-sensitive
6-bit	~1.2 GB	Approaching FP16 quality	High-fidelity
8-bit	~1.5 GB	Near-lossless reference	Fidelity-critical work

(Current variant — 2bit — is bolded.)

Variants in this family

(Showing 14 sibling variants under majentik/gemma-4-e2b-it-*. The current variant — RotorQuant-MLX-2bit — is bolded.)

Variant	Runtime	Approx size	Use case
RotorQuant	runtime modifier	n/a	KV-cache root (weight-agnostic)
RotorQuant-GGUF-IQ4_XS	llama.cpp	~1.7 GB	Lossy 4-bit, low-RAM CPU/edge
RotorQuant-GGUF-Q2_K	llama.cpp	~1.2 GB	Lossy, low-RAM CPU/edge
RotorQuant-GGUF-Q3_K_M	llama.cpp	~1.6 GB	Smaller 3-bit, CPU-friendly
RotorQuant-GGUF-Q4_K_M	llama.cpp	~2.2 GB	Balanced default
RotorQuant-GGUF-Q5_K_M	llama.cpp	~2.6 GB	Higher fidelity, more RAM
RotorQuant-GGUF-Q8_0	llama.cpp	~4.2 GB	Near-lossless reference
RotorQuant-MLX-2bit	mlx-lm	~655 MB	Apple Silicon, smallest
RotorQuant-MLX-4bit	mlx-lm	~1.2 GB	Apple Silicon balanced
RotorQuant-MLX-8bit	mlx-lm	~2.4 GB	Apple Silicon reference
TurboQuant	runtime modifier	n/a	KV-cache root (weight-agnostic)
TurboQuant-MLX-2bit	mlx-lm	~655 MB	Apple Silicon, smallest
TurboQuant-MLX-4bit	mlx-lm	~1.2 GB	Apple Silicon balanced
TurboQuant-MLX-8bit	mlx-lm	~2.4 GB	Apple Silicon reference