llmrun
z-labGemma 4DFlashDraftModel

Gemma 4 31B IT DFlash — Hardware Requirements & GPU Compatibility

Chat

Gemma 4 31B IT DFlash is a 31B-parameter open language model from z-lab in the Gemma 4 family. It supports a context window of up to 262,144 tokens. At Q4_K_M it needs about 18.93 GB of VRAM — see which GPUs and Macs can run it below.

31B params · dense262K contextApache 2.018.5K downloads111 likes
A

Runs well on a 24 GB GPU

Runs from16 GB Q3_K_M · 32KBest on24 GB Q4_K_M · ~35 tok/s · 256KMacs from18 GB Q2_K · ~8 tok/s

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Gemma 4 31B IT DFlash

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Recommended

vLLM
one command · needs vLLM installed
$vllm serve z-lab/gemma-4-31B-it-DFlash --max-model-len 16384

vLLM repo with its own quantisation

Download: Weights on Hugging Face

install, OS, other quants

Other ways

each opens step-by-step

Run in cloud

Fits on RTX 3090 (24 GB) (4 GB headroom) · Q4_K_M

Generation speed
~32 tok/s
generation speed
Cost per 1M output tokens
$1.87
per 1M output tokens
Deployon Vast.aiBest optionfrom $0.22/hr (opens in new tab)
Compare GPUs →
or

How Much VRAM Does Gemma 4 31B IT DFlash Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KOn RTX 4090 (reference)
13.5 GB13.9 GB2-bit · K-quant — 2-bit quantization with K-quant improvementsFits: 256K context · ~49 tok/s
15.4 GB15.9 GB3-bit medium — 3-bit medium quantizationFits: 256K context · ~42 tok/s
18.9 GB19.3 GB4-bit medium — 4-bit medium quantization — most popular sweet spotFits: 256K context · ~35 tok/s
22.4 GB22.8 GB5-bit medium — 5-bit medium quantization — good quality/size tradeoffFits: 64K context · ~29 tok/s
25.9 GB26.3 GB6-bit, very good — 6-bit quantization, very good qualityDoes not fit: needs 26.1 GB
31.3 GB31.7 GB8-bit, near-lossless — 8-bit quantization, near-losslessDoes not fit: needs 31.5 GB
62.3 GB62.7 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingDoes not fit: needs 62.5 GB
est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

19.1GBat 16K context

Weights
18.6 GB
KV cache
0.2 GB
Overhead
0.3 GB

Fits RTX 4090 (reference) · 4.9 GB free · ~35 tok/s

≥50 tok/s needs ≥1.46 TB/s · RTX 5090 ~62

Which Hardware Can Run Gemma 4 31B IT DFlash?

Q4_K_M · 18.9 GB

Gemma 4 31B IT DFlash (Q4_K_M) requires 18.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 25+ GB is recommended. Using the full 262K context window can add up to 3.5 GB, bringing total usage to 22.4 GB. 8 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 122 devices with unified memory can run Gemma 4 31B IT DFlash, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Mini M5 Pro (24 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M
8–12 GB · RTX 4060, RTX 3060 12GBDoesn't fit — smallest quant needs 13.7 GB8 GB tier →12 GB tier →
16 GB · RTX 5060 Ti 16GBQ3_K_M · 15.6 GB · ~19 tok/s · 32K16 GB tier →
24 GB · RTX 4090referenceQ4_K_M · 19.1 GB · ~35 tok/s · 256K24 GB tier →
Mac 32 GB · Mac Mini M4Q4_K_M · 19.1 GB · ~4 tok/s · 256KMacs →
Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 19.1 GB · ~20 tok/s · 256KMacs →

Alternatives to Gemma 4 31B IT DFlash

Gemma 4 family · 35 models →

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
Hy Memory 30B A3BTencent—19.1 GB~184same size
Llm Jp 4.1 32B A3b Thinkingllm-jp—20.1 GB~188same size
Llm Jp 4.1 33B Thinkingllm-jp—22.5 GB~35same size
Agnes 3.0 FlashAgnes-AI—22.1 GB~35same size
Muse Glimmer 30BMeta—19.6 GB~36same size
NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA—19.3 GB~197same size

Specifications

Publisher
z-lab
Family
Gemma 4
Parameters
31B
Architecture
DFlashDraftModel
Context Length
262,144 tokens
Vocabulary Size
262,144
Release Date
2026-04-30
License
Apache 2.0

Resources

Frequently Asked Questions

How much VRAM does Gemma 4 31B IT DFlash need?

Gemma 4 31B IT DFlash requires 18.9 GB of VRAM at Q4_K_M, or 62.3 GB at BF16. Full 262K context adds up to 3.5 GB (22.4 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 31B × 4.8 bits ÷ 8 = 18.6 GB

KV Cache + Overhead ≈ 0.3 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 3.8 GB (at full 262K context)

VRAM usage by quantization

18.9 GB
22.4 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Gemma 4 31B IT DFlash?

Yes, at Q5_K_M (22.4 GB) or lower. Higher quantizations like Q6_K (25.9 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Gemma 4 31B IT DFlash?

For Gemma 4 31B IT DFlash, Q4_K_M (18.9 GB) offers the best balance of quality and VRAM usage. Q5_K_M (22.4 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 13.5 GB.

VRAM requirement by quantization

Q2_K
13.5 GB
Q4_K_M ★
18.9 GB
Q5_K_M
22.4 GB
Q6_K
25.9 GB
Q8_0
31.3 GB
BF16
62.3 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Gemma 4 31B IT DFlash on a Mac?

Yes — MacBook Pro 14" M3 Pro (18 GB) and 108 other Macs can run Gemma 4 31B IT DFlash, the smallest at Q2_K. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 18.9 GB of usable unified memory (RAM minus macOS overhead), from Mac Mini M5 Pro (24 GB).

Can I run Gemma 4 31B IT DFlash locally?

Yes — Gemma 4 31B IT DFlash can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Gemma 4 31B IT DFlash?

At Q4_K_M, Gemma 4 31B IT DFlash can reach ~254 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~35 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 18.9 × 0.65 = ~275 tok/s

Estimated speed at Q4_K_M (18.9 GB)

~275 tok/s
~35 tok/s
~275 tok/s
~254 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Gemma 4 31B IT DFlash?

At Q4_K_M, the download is about 18.60 GB. The full-precision BF16 version is 62.00 GB. The smallest option (Q2_K) is 13.18 GB.

Which GPUs can run Gemma 4 31B IT DFlash?

8 consumer GPUs can run Gemma 4 31B IT DFlash at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XT, AMD Radeon RX 7900 XTX. 1 GPU has plenty of headroom for comfortable inference.

Which devices can run Gemma 4 31B IT DFlash?

122 devices with unified memory can run Gemma 4 31B IT DFlash at Q4_K_M (18.9 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.