跳转到主要内容

Gemma 3 4B Instruct

GoogleGemma开源权重Gemma · 商用许可

描述

Gemma 3 4B is a 4-billion-parameter vision-language model from Google, handling text and image input and generating text output. It features a 128K context window, multilingual support, and open weights. Suitable for question answering, summarization, reasoning, and image understanding tasks.

发布日期
2025-03-12
参数规模
4.0B
上下文长度
131K
支持模态
image, text

能力雷达图

14
general
6
coding
22
reasoning
17
science
17
agents
70
multimodal

排行榜排名

领域#排名分数来源
代码能力榜533
5.0
AA
通用能力榜550
14.0
AA
多模态榜70
36.0
LS
科学能力543
14.0
AA

基准测试分数 (LLM Stats)

(LLM Stats (zeroeval))

Biology

GPQANYU + Cohere + Anthropic (2023)30.8%自报

Code

HumanEvalOpenAI (2021)71.3%自报
LiveCodeBench12.6%自报

Factuality

FACTS Grounding70.1%自报
SimpleQA4.0%自报

Finance

MMLU-Pro43.6%自报

General

IFEvalGoogle Research (2023)90.2%自报
Natural2Code70.3%自报
MBPP0.63 / 100自报
Global-MMLU-Lite54.5%自报
MMMU (val)48.8%自报
BIG-Bench Extra Hard11.0%自报

Image To Text

DocVQADocVQA (2020)75.8%自报
VQAv2 (val)62.4%自报
TextVQA57.8%自报

Language

BIG-Bench Hard72.2%自报
WMT24++46.8%自报
ECLeKTic4.6%自报

Math

GSM8k89.2%自报
MATH75.6%自报
MathVista-Mini50.0%自报
HiddenMath43.0%自报

Multimodal

AI2D74.8%自报
ChartQAMasry et al. (2022)68.8%自报
InfoVQA50.0%自报

Reasoning

Bird-SQL (dev)36.3%自报

AA 评测指数

(Artificial Analysis)
Math Index(Artificial Analysis)
12.7
Coding Index(Artificial Analysis)
2.7
Intelligence Index(Artificial Analysis)
1.0
Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))
0.8
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))
0.4
Gpqa(NYU + Cohere + Anthropic (2023))
0.3
Ifbench(Google Research (2023))
0.3
Aime 25(MAA (Mathematical Association of America))
0.1
Livecodebench(UC Berkeley + MIT + Cornell (2024))
0.1
Scicode(UIUC + Argonne National Lab (2024))
0.1
Lcr(Artificial Analysis)
0.1
Aime(MAA (Mathematical Association of America))
0.1
Hle(Center for AI Safety + Scale AI (2025))
0.1
Tau2(Sierra + U Toronto + Vector Institute (2025))
0.0
Terminalbench Hard(Stanford × Laude Institute (2026))
0.0
Tau Banking
0.0
Terminalbench V2 1
0.0

LLM Stats 分类评分

(LLM Stats (zeroeval))
Structured Output
90
Instruction Following
90
Image To Text
70
Grounding
70
Math
60
Multimodal
60
Vision
60
Reasoning
50
General
50
Healthcare
50
Legal
40
Language
40
Factuality
40
Finance
40
Code
40
Physics
30
Biology
30
Chemistry
30

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0.0
首Token延迟0.00s
首回答延迟0.00s

供应商价格排行

供应商价格排行

2 个供应商

最便宜: OpenRouter最贵: Kilo Gateway
供应商输入输出
1OpenRouter最便宜
$0.05
$0.1
2Kilo Gateway
$0.05
$0.1

比较该模型在不同 API 供应商之间的定价。

外部链接