跳转到主要内容

Gemma 4 31B (Reasoning)

GoogleGemmaOpen WeightApache 2.0 · Commercial OK

描述

Gemma 4 31B is Google DeepMind's flagship dense multimodal model with 31 billion parameters and a 256K context window. Ranks #3 among open models on Arena AI. Built from the same research as Gemini 3, it features Per-Layer Embeddings, Shared KV Cache, alternating sliding-window and global attention, and variable aspect ratio vision encoding. Achieves an estimated LMArena text score of 1452.

发布日期
2026-04-02
参数规模
30.7B
上下文长度
262K
支持模态
image, text, video

能力雷达图

36
general
39
coding
86
reasoning
58
science估算
90
agents
70
multimodal

Science 在缺少专门科学评测时使用推理能力代理估算。

排行榜排名

领域#排名分数来源
智能体与工具1
86.0
LS
代码能力榜63
68.0
AA
通用能力榜90
69.0
AA
科学能力51
71.0
AA

基准测试分数 (LLM Stats)

Agents

t2-bench86.4%自报

Biology

GPQA84.3%自报

Finance

MMLU-Pro85.2%自报

General

MMMLU88.4%自报
LiveCodeBench v680.0%自报
MMMU-Pro76.9%自报
BIG-Bench Extra Hard74.4%自报
MRCR v266.4%自报

Healthcare

MedXpertQA61.3%自报

Math

AIME 202689.2%自报
MathVision85.6%自报
Humanity's Last Exam26.5%自报

AA 评测指数

Intelligence Index
39.2
Coding Index
38.7
Gpqa
0.9
Ifbench
0.8
Lcr
0.6
Tau2
0.6
Scicode
0.4
Terminalbench Hard
0.4
Hle
0.2

LLM Stats 分类评分

Tool Calling
90
Agents
90
Finance
90
Legal
90
Biology
80
Chemistry
80
General
80
Language
80
Physics
80
Math
70
Multimodal
70
Reasoning
70
Vision
60
Healthcare
60
Long Context
60

定价

输入价格免费
输出价格免费
混合价格(3:1)免费

速度

Tokens/秒34.9 tokens/s
首Token延迟1.02s
首回答延迟58.33s

可用提供商

(LS 内部计价单位)
提供商输入价格输出价格
Novita140K400K

外部链接