跳转到主要内容

Gemini 2.5 Flash-Lite (Non-reasoning)

GoogleGeminiOpen WeightCreative Commons Attribution 4.0 License

描述

Gemini 2.5 Flash-Lite is a model developed by Google DeepMind, designed to handle various tasks including reasoning, science, mathematics, code generation, and more. It features advanced capabilities in multilingual performance and long context understanding. It is optimized for low latency use cases, supporting multimodal input with a 1 million-token context length.

发布日期
2025-06-17
参数规模
上下文长度
1.0M
支持模态
audio, file, image, text, video

能力雷达图

29
general
20
coding
49
reasoning
28
science估算
0
agents
89
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
音频能力23
71.0
AA
代码能力榜312
23.0
AA
通用能力榜361
31.0
AA
数学推理179
51.0
AA
科学能力379
27.0
AA

基准测试分数 (LLM Stats)

Biology

GPQA64.6%自报

Code

LiveCodeBench33.7%自报
SWE-Bench Verified31.6%自报
Aider-Polyglot26.7%自报

Factuality

FACTS Grounding84.1%自报
SimpleQA10.7%自报

General

Global-MMLU-Lite81.1%自报
MMMU72.9%自报
Vibe-Eval51.3%自报
MRCR v216.6%自报
Arc2.5%自报

Math

AIME 202549.8%自报
Humanity's Last Exam5.1%自报

AA 评测指数

Math Index
35.3
Intelligence Index
12.7
Coding Index
7.4
Math 500
0.9
Mmlu Pro
0.7
Aime
0.5
Gpqa
0.5
Livecodebench
0.4
Aime 25
0.4
Ifbench
0.3
Lcr
0.3
Tau2
0.2
Scicode
0.2
Hle
0.0
Terminalbench Hard
0.0

LLM Stats 分类评分

Grounding
80
Language
80
Healthcare
70
Biology
60
Chemistry
60
Multimodal
60
Physics
60
Factuality
50
Reasoning
50
Vision
40
General
40
Code
30
Frontend Development
30
Math
30
Long Context
20

定价

输入价格$0.1 / 1M tokens
输出价格$0.4 / 1M tokens
混合价格(3:1)$0.175 / 1M tokens

速度

Tokens/秒267.1 tokens/s
首Token延迟0.53s
首回答延迟0.53s

可用提供商

(LS 内部计价单位)

暂无提供商数据

外部链接