Gemma 4 31B

Name: Gemma 4 31B
Price: 0.16 USD
Author: Google

Google · Gemma 4 · released 2026-04-02 · data checked 2026-11-01

Gemma 4 31B is a 30.7B-parameter dense model from Google (United States), released 2026-04-02 under the Apache-2.0 license. It scores 70.1 on the Small LM Index, #8 of 34 medium models. It needs about 21.9 GB of memory at 4-bit with an 8K-token context and costs $0.16 per million tokens through the cheapest API provider.

Small LM Index

70.1

#8 of 34 scored medium models

Likely 68.7–71.5 · from 6 of 6 core benchmarks

Parameters

30.7B

Dense

Price per 1M tokens

$0.16

$0.10 in / $0.34 out · blended 3:1

Context window

256K

tokens

Benchmarks

Each bar shows this model's score. The tick marks the best medium model on that benchmark. Sources: 7 reported by Google, 2 measured by another lab. How we label sources.

MMLU-Pro Lab-reported Knowledge85.2

Class best: Qwen3.6 27B 86.2

GPQA Diamond Lab-reported Knowledge84.3

Class best: Nanbeige4.3 33B 89.5

Humanity's Last Exam Lab-reported Reasoning19.5

Class best: Qwen3.8 27B 30.8

AIME Lab-reported Math · AIME 2026, no tools89.2

Class best: Ternary Bonsai 2 27B 95.8

LiveCodeBench Lab-reported Coding · LiveCodeBench v680.0

Class best: Qwen3.8 27B 90.3

SWE-bench Verified Other lab Coding · Reported by Alibaba (Qwen3.6 card)52.0

Class best: Qwen3.6 27B 77.2

Terminal-Bench Other lab Agentic · Terminal-Bench 2.0, reported by Alibaba42.9

Class best: Nanbeige4.3 33B 80.9

τ²-Bench Lab-reported Agentic76.9

Class best: MiniCPM-V 4.7 35B-A3B 84.7

MMMU-Pro Lab-reported Vision76.9

Specifications

Lab: Google (US · United States)
Architecture: Dense transformer
Total parameters: 30.7B
Active parameters: All (dense)
Size class: Medium
Context window: 256K tokens
Input modalities: text, image
Reasoning: Hybrid (thinking on/off)
License: Apache-2.0
Openness: Open weights
Memory to run: ~21.9 GB at 4-bit, 8K context (more sizes)
Hugging Face: 8M downloads (30 days) · 4.3K likes
API providers: 14 on OpenRouter
API price: $0.10 in / $0.34 out per 1M (OpenRouter, 2026-11-03)

Weights on Hugging Face ↗Try on OpenRouter ↗

Compare with

Gemma 4 31B vs Muse Glimmer 30B (29.6B, index 72.6)
Gemma 4 31B vs Kanana-2 30B-A3B Thinking (30.7B, index 50.7)
Gemma 4 31B vs Qwen3 30B-A3B Thinking 2507 (30.5B, index 59.3)
Gemma 4 31B vs GLM-4.7-Flash (31B, index 66.1)
Gemma 4 31B vs Granite 4.2 30B (30B, index 67.0)

Ranked in

Sources

https://huggingface.co/google/gemma-4-31B-it

Something wrong here? Report it.

Run it locally

Uses ggml-org/gemma-4-31B-it-GGUF (Q4_0). Swap the tag for another size: BF16, Q4_0, Q8_0.

Ollama

ollama run hf.co/ggml-org/gemma-4-31B-it-GGUF:Q4_0

llama.cpp

llama-server -hf ggml-org/gemma-4-31B-it-GGUF:Q4_0

Then open http://localhost:8080 for a chat window.

MLX (Mac)

pip install -U mlx-lm
mlx_lm.chat --model mlx-community/gemma-4-31b-it-4bit

Text only; for images use mlx-vlm.

vLLM (GPU server)

vllm serve google/gemma-4-31B-it

Full-precision weights, for a server GPU.

Open in LM Studio

Memory needed: weights, context cache and runtime buffers.

Quantization	4K context	32K context	256K context
16-bit F16	64 GB	69 GB	105 GB
8-bit Q8_0	35 GB	40 GB	77 GB
4-bit Q4_K_M	21 GB	26 GB	63 GB

Check it against your GPU, Mac or phone →