Gemma 4 31B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
Gemma 4's flagship open-weight model. Full dense 31B with text + image + audio + video multimodal capability — the only sub-50B open model with native video understanding. Scores 92% on MMLU and 86.4% on HumanEval, rivalling GPT-4o on most benchmarks. Needs ~18–20 GB at Q4_K_M (fits on RTX 3090/4090 or 24 GB Mac). Apache 2.0 licensed.
Gemma 4 31B needs about 20 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
VRAM and speed by quantization
Quoted against NVIDIA RTX 4090 (24 GB). Includes the KV cache at 8K context, so it reads higher than the headline figure.
| Quant | Memory | VRAM | Speed (est.) | Fit |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 12.9 GB | ~56 tok/s | Fits | |
| Q3_K_M 3.41 bpw | 15.9 GB | ~46 tok/s | Fits | |
| Q4_K_M 4.83 bpw | 21.4 GB | ~34 tok/s | Fits | |
| Q5_K_M 5.67 bpw | 24.7 GB | ~4 tok/s | Offload | |
| Q6_K 6.56 bpw | 28.1 GB | ~4 tok/s | Offload | |
| Q8_0 8.50 bpw | 35.7 GB | ~3 tok/s | Offload | |
| F16 16.00 bpw | 64.7 GB | — | Too big |
Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. Gemma 4 31B VRAM calculator →
Get Gemma 4 31B running
The cheapest catalogued GPU that runs Gemma 4 31B is the AMD Radeon RX 7900 XTX (24 GB).
or compare on Vast.ai from $0.35/hr (typical low · varies)
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run Gemma 4 31B
Install Ollama, then run:
ollama run gemma4:31bSpecifications
Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.
- Parameters
- 31 Billion
- Context window
- 256,000
- Architecture
- Dense Transformer + Full Multimodal Encoder (image + audio + video)
- Provider
- Licence
- Apache 2.0
- Practical deployment
- Single consumer GPU
- Specified at
- Q4_K_M
- System RAM
- 40 GB
- Record updated
- 2026-03-31
Commercial use permitted. No usage restrictions beyond attribution.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
| Benchmark | Score | Provenance |
|---|---|---|
| MMLU | 92.1 / 100 % | reported |
| HumanEval | 86.4 / 100 % | reported |
| MATH | 80.3 / 100 % | reported |
| MMMU (Multimodal) | 74.8 / 100 % | reported |
Can I run Gemma 4 31B on my GPU?
- Gemma 4 31B on AMD Radeon RX 7800 XT
- Gemma 4 31B on AMD Radeon RX 7900 XT
- Gemma 4 31B on AMD Radeon RX 7900 XTX
- Gemma 4 31B on AMD Radeon RX 9060 XT 16GB
- Gemma 4 31B on AMD Radeon RX 9060 XT 8GB
- Gemma 4 31B on AMD Radeon RX 9070
- Gemma 4 31B on AMD Radeon RX 9070 XT
- Gemma 4 31B on Apple M1
- Gemma 4 31B on Apple M1 Pro
- Gemma 4 31B on Apple M2
- Gemma 4 31B on Apple M2 Pro
- Gemma 4 31B on Apple M3
- Gemma 4 31B on Apple M3 Pro
- Gemma 4 31B on Apple M4
- Gemma 4 31B on Apple M4 Pro
- Gemma 4 31B on Apple M5
- Gemma 4 31B on Intel Arc B570
- Gemma 4 31B on Intel Arc B580
- Gemma 4 31B on NVIDIA GeForce RTX 3060 (12GB)
- Gemma 4 31B on NVIDIA GeForce RTX 3070
- Gemma 4 31B on NVIDIA GeForce RTX 3070 Ti
- Gemma 4 31B on NVIDIA GeForce RTX 3080 (10GB)
- Gemma 4 31B on NVIDIA GeForce RTX 3090
- Gemma 4 31B on NVIDIA GeForce RTX 4060
- Gemma 4 31B on NVIDIA GeForce RTX 4060 Ti 16GB
- Gemma 4 31B on NVIDIA GeForce RTX 4070
- Gemma 4 31B on NVIDIA GeForce RTX 4070 Super
- Gemma 4 31B on NVIDIA GeForce RTX 4070 Ti
- Gemma 4 31B on NVIDIA GeForce RTX 4070 Ti Super
- Gemma 4 31B on NVIDIA GeForce RTX 4080
- Gemma 4 31B on NVIDIA GeForce RTX 4080 Super
- Gemma 4 31B on NVIDIA GeForce RTX 4090
- Gemma 4 31B on NVIDIA GeForce RTX 5060
- Gemma 4 31B on NVIDIA GeForce RTX 5060 Ti 16GB
- Gemma 4 31B on NVIDIA GeForce RTX 5060 Ti 8GB
- Gemma 4 31B on NVIDIA GeForce RTX 5070
- Gemma 4 31B on NVIDIA GeForce RTX 5070 Ti
- Gemma 4 31B on NVIDIA GeForce RTX 5080
- Gemma 4 31B on NVIDIA GeForce RTX 5090
Other Gemma 4 sizes
Gemma 4 31B — frequently asked questions
How much VRAM does Gemma 4 31B need?
About 20 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does Gemma 4 31B run on an RTX 4090 (24 GB)?
Yes. Gemma 4 31B needs about 20 GB at Q4_K_M, inside a 24 GB card, at an estimated 34 tokens/sec.
How do I run Gemma 4 31B locally?
Install Ollama and run `ollama run gemma4:31b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.
What other sizes does Gemma 4 come in?
Gemma 4 E2B (4 GB), Gemma 4 E4B (6 GB), Gemma 4 26B-A4B (16 GB), Gemma 4 31B (20 GB), Gemma 4 12B (Unified) (8 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.