Model record
Gemma 4 31B
Google DeepMind · Apache 2.0 · verified
What this model is
The architecture is dense. Every parameter is part of the active path, so memory and speed scale with the full model rather than a smaller expert subset.
Memory and quantization
Decimal GB. Sizes combine the published Unsloth GGUF weight file with the multimodal projector required for image input.
| Quantization | Published files | 8K estimate | Smallest comfortable GPU | Smallest comfortable Mac | Context at smallest GPU / Mac | Speed estimate | Run commands |
|---|---|---|---|---|---|---|---|
| Q4_K_M | 19.52 GB | 22.20 GB | NVIDIA GeForce RTX 5090 32GB · 32 GB VRAM | Apple Mac mini M4 Pro 48GB · 48 GB unified memory | 76,562 / 188,170 tokens | Estimate · Server relative class | ollama run gemma4:31b-it-q4_K_M lms get "Gemma 4 31B"@q4_k_m |
| Q8_0 | 33.84 GB | 36.51 GB | NVIDIA RTX 6000 Ada 48GB · 48 GB VRAM | Apple Mac mini M4 Pro 48GB · 48 GB unified memory | 88,337 / 88,337 tokens | Estimate · Server relative class | ollama run gemma4:31b-it-q8_0 lms get "Gemma 4 31B"@q8_0 |
Estimate formula: published weight files + 1.1744 GB per 8K tokens × context / 8K + 1.5 GB runtime overhead. Sizes are decimal GB, not GiB. At 8K context, the smallest listed build is about 22.2 GB before OS and application reserve.
Conservative FP16 estimate from the public config. Ten of 60 layers use global attention and local layers use a 1024-token sliding window.
Ollama tags were checked against the linked Ollama library pages on 2026-11-02. LM Studio's lms get suffix selects a file when the search results expose that exact quantization; confirm the local key with lms ls.
KV method source: https://huggingface.co/google/gemma-4-31B-it/raw/main/config.json
Speed heuristic
Not benchmarked here. Dense 30.7B inference activates the full model on every token.
Active path: 30.7B parameters. This is a relative speed signal only; it is not a tokens/sec benchmark.
Runtimes
- Ollama
- llama.cpp
- MLX
ollama run gemma4:31blms get "Gemma 4 31B"@q4_k_m
lms load <model-key-from-lms-ls> --gpu max --context-length 8192License restrictions
Apache 2.0; the separate Gemma Prohibited Use Policy and intended-use statement still apply.
This is a plain-language summary, not legal advice. Read the linked license before commercial deployment or redistribution.
Known issues
- A Q4 build plus KV cache is a workstation-class workload, not a typical 16 GB laptop fit.
- The GGUF package needs the multimodal projector for image input.
If it does not fit, try these smaller records
These alternatives have a lower recommended memory target in the current catalog. Open each page to compare the quantization-specific fit before downloading.
Qwen3.8 27B
Dense · starting at 20 GB
gpt-oss-20b
MoE · starting at 14 GB
Qwen3 8B
Dense · starting at 8 GB
DeepSeek R1 8B
Dense · starting at 8 GB
FAQ
Model-specific questions
What is the smallest local file for Gemma 4 31B?
The smallest published weight package in this record is Q4_K_M at 19.52 GB. The 8K runtime estimate adds 1.17 GB of KV cache and 1.5 GB overhead, for about 22.2 GB before operating-system reserve.
How much memory do the published quantizations need at 8K?
Q4_K_M about 22.2 GB at 8K; Q8_0 about 36.5 GB at 8K. These are calculated estimates using the published weight sizes and the model's KV-cache assumption, not live memory measurements.
Which GPU and Mac configurations are comfortable for Gemma 4 31B?
Against the current catalog, the smallest configurations that clear the 20% headroom target are: Q4_K_M: NVIDIA GeForce RTX 5090 32GB; Q8_0: NVIDIA RTX 6000 Ada 48GB. A larger configuration may still be preferable for longer context or faster memory bandwidth.
What is the right Ollama or LM Studio command for Gemma 4 31B?
Use the command attached to the specific quantization in the table. Ollama tags are taken from the checked Ollama library pages. For LM Studio, use lms get with the displayed quantization suffix, then load the resulting local key from lms ls with --gpu max and the desired context length.
What should I run instead if Gemma 4 31B does not fit?
Smaller catalog alternatives include Qwen3.8 27B, gpt-oss-20b, Qwen3 8B, DeepSeek R1 8B. Compare their quantization tables and minimum memory before choosing; a smaller parameter count does not always mean a proportionally smaller download.
Sources
- https://huggingface.co/google/gemma-4-31B-it
- https://huggingface.co/unsloth/gemma-4-31B-it-GGUF
- https://ollama.com/library/gemma4
- https://huggingface.co/google/gemma-4-31B-it/raw/main/config.json
This page is an independent compatibility guide, not official documentation. Verify the primary source before downloading or deploying a model.