Which needs less VRAM, Gemma 4 31B or gemma-3-27b-it?

gemma-3-27b-it: about 20.1 GB at Q4_K_M, against 22.8 GB for Gemma 4 31B. At Q8_0 it's 34.5 GB vs 39.2 GB.

Which is faster locally, Gemma 4 31B or gemma-3-27b-it?

Decode speed is limited by how many bytes are read per token. gemma-3-27b-it reads 27.4B parameters per token and Gemma 4 31B reads 31.3B, so gemma-3-27b-it should be faster on the same machine. Our estimate on the GeForce RTX 5090 at Q4 is 65–92 tok/s vs 57–80 tok/s. These are estimates, not benchmarks.

What is the cheapest GPU for Gemma 4 31B and gemma-3-27b-it?

At Q4: Intel Arc Pro B70 32GB ($949) for Gemma 4 31B; Intel Arc Pro B70 32GB ($949) for gemma-3-27b-it. Prices are dated list-price snapshots.

Can a Mac run Gemma 4 31B and gemma-3-27b-it?

The cheapest Mac or mini PC we track for Gemma 4 31B is the Mac mini M5 Pro 64GB ($3,199). For gemma-3-27b-it it's the Mac mini M6 32GB ($1,799).

Which is better, Gemma 4 31B or gemma-3-27b-it?

We don't score answer quality. This comparison covers memory, context cache, hardware and speed. Pick the one that fits your hardware, then test both on your own prompts.

Gemma 4 31B vs gemma-3-27b-it: VRAM and speed for running locally

Side by side: how much memory each model needs at Q4, Q8 and FP16, how the KV cache grows with context, the cheapest GPU and Mac that run each, and estimated speed on 11 common machines.

gemma-3-27b-it is lighter: about 20.1 GB at Q4 against 22.8 GB for Gemma 4 31B, so it runs on 25 tracked devices vs 16. It should also be faster on the same hardware, since it reads 27.4B parameters per token against 31.3B. Which one gives better answers depends on your task, so check benchmarks for that; this page covers what it takes to run them.

Gemma 4 31B vs gemma-3-27b-it: specs and memory
	Gemma 4 31B	gemma-3-27b-it
Maker	google	google
License	apache-2.0	gemma
Type	Dense	Dense
Total parameters	31.3B	27.4B
Active per token	31.3B all	27.4B all
Memory at Q4_K_M	22.8 GB	20.1 GB
Memory at Q8_0	39.2 GB	34.5 GB
Memory at FP16	72.9 GB	64.1 GB
KV cache growth	not modeled	not modeled
Memory at Q4, 32k context	cache not modeled	cache not modeled
Cheapest GPU at Q4	Intel Arc Pro B70 32GB $949	Intel Arc Pro B70 32GB $949
Cheapest Mac / mini PC at Q4	Mac mini M5 Pro 64GB $3,199	Mac mini M6 32GB $1,799
Cheapest GPU at Q8	NVIDIA RTX 6000 Ada 48GB $6,799	NVIDIA RTX 6000 Ada 48GB $6,799
Tracked devices that run it at Q4	16 of 82	25 of 82

Estimated speed on the same hardware

Decode tokens per second at Q4_K_M from memory bandwidth (how we estimate). MoE figures are upper bounds ("up to"). Estimates, not benchmarks.

Estimated tokens per second, Gemma 4 31B vs gemma-3-27b-it
Device	Gemma 4 31B	gemma-3-27b-it
GeForce RTX 3060 12GB12 GB usable · 360 GB/s · $329	doesn’t fit	doesn’t fit
GeForce RTX 4060 Ti 16GB16 GB usable · 288 GB/s · $499	doesn’t fit	doesn’t fit
GeForce RTX 508016 GB usable · 960 GB/s · $999	doesn’t fit	doesn’t fit
GeForce RTX 309024 GB usable · 936 GB/s · $1,499	tight fit	34–48
Radeon RX 7900 XTX24 GB usable · 960 GB/s · $999	tight fit	35–49
GeForce RTX 509032 GB usable · 1792 GB/s · $1,999	57–80	65–92
Mac mini M6 24GB18 GB usable · 170 GB/s · $1,299	doesn’t fit	doesn’t fit
Mac mini M6 32GB24 GB usable · 170 GB/s · $1,799	tight fit	5–8
Mac mini M5 Pro 64GB48 GB usable · 307 GB/s · $3,199	7–13	8–15
Framework Desktop (Ryzen AI Max+ 395) 128GB96 GB usable · 256 GB/s · $3,449	6–11	7–12
MacBook Pro M5 Max 128GB96 GB usable · 614 GB/s · $6,699	15–26	17–30

Hardware that runs only one of them

Run gemma-3-27b-it but not Gemma 4 31B: Radeon RX 7900 XTX, GeForce RTX 3090, GeForce RTX 4090, Mac mini M6 32GB, GeForce RTX 3090 Ti, MacBook Pro M6 32GB.

Current priced devices, cheapest first, at Q4 with ~4k context.

Quick answers

Which needs less VRAM, Gemma 4 31B or gemma-3-27b-it?: gemma-3-27b-it: about 20.1 GB at Q4_K_M, against 22.8 GB for Gemma 4 31B. At Q8_0 it's 34.5 GB vs 39.2 GB.
Which is faster locally, Gemma 4 31B or gemma-3-27b-it?: Decode speed is limited by how many bytes are read per token. gemma-3-27b-it reads 27.4B parameters per token and Gemma 4 31B reads 31.3B, so gemma-3-27b-it should be faster on the same machine. Our estimate on the GeForce RTX 5090 at Q4 is 65–92 tok/s vs 57–80 tok/s. These are estimates, not benchmarks.
What is the cheapest GPU for Gemma 4 31B and gemma-3-27b-it?: At Q4: Intel Arc Pro B70 32GB ($949) for Gemma 4 31B; Intel Arc Pro B70 32GB ($949) for gemma-3-27b-it. Prices are dated list-price snapshots.
Can a Mac run Gemma 4 31B and gemma-3-27b-it?: The cheapest Mac or mini PC we track for Gemma 4 31B is the Mac mini M5 Pro 64GB ($3,199). For gemma-3-27b-it it's the Mac mini M6 32GB ($1,799).
Which is better, Gemma 4 31B or gemma-3-27b-it?: We don't score answer quality. This comparison covers memory, context cache, hardware and speed. Pick the one that fits your hardware, then test both on your own prompts.

Related comparisons

Qwen3.8-27B vs Gemma 4 31B20.4 vs 22.8 GB
Mistral-Small-3.2-24B-Instruct-2506 vs gemma-3-27b-it17.7 vs 20.1 GB

All comparisons · MoE vs dense · Q4 vs Q8 vs FP16 · Context length and VRAM · VRAM calculator

Gemma 4 31B vs gemma-3-27b-it: VRAM and speed for running locally

Estimated speed on the same hardware

Hardware that runs only one of them

Quick answers

Related comparisons

New models, and what they need to run