Mistral Small (Sep '24)
MistralMistralOpen WeightApache 2.0 · Commercial OK
描述
Mistral Small 4 is Mistral AI's hybrid small model that unifies instruct, reasoning, multimodal understanding, and coding capabilities in a single efficient system. It uses a 119B-parameter Mixture-of-Experts architecture with native image input, a 256k context window, and configurable reasoning depth for general chat, coding agents, and long-context reasoning workloads.
发布日期
2024-09-17
参数规模
119.0B
上下文长度
—
支持模态
image, text
能力雷达图
22
general
14
coding
28
reasoning
23
science估算
0
agents
60
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
基准测试分数 (LLM Stats)
Biology
GPQA
71.2%自报
Code
LiveCodeBench
63.6%自报
Creativity
Arena Hard
58.3%自报
Finance
MMLU-Pro
78.0%自报
General
MMMU-Pro
60.0%自报
IFBench
48.0%自报
Language
COLLIE
62.9%自报
Long Context
AA-LCR
71.2%自报
Math
AIME 2025
83.8%自报
AA 评测指数
Intelligence Index10.2
Math 5000.6
Mmlu Pro0.5
Gpqa0.4
Scicode0.2
Livecodebench0.1
Aime0.1
Hle0.0
LLM Stats 分类评分
Finance80
Healthcare80
Legal80
Math80
Biology70
Chemistry70
Language70
Long Context70
Physics70
Reasoning70
Vision60
Writing60
Code60
Creativity60
General60
Multimodal60
Instruction Following50
定价
输入价格$0.2 / 1M tokens
输出价格$0.6 / 1M tokens
混合价格(3:1)$0.3 / 1M tokens
速度
Tokens/秒155.0 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)| 提供商 | 输入价格 | 输出价格 |
|---|---|---|
| Mistral AI | 150K | 600K |