DeepSeek-V2.5 (Dec '24)
DeepSeekDeepSeekOpen Weightdeepseek
描述
DeepSeek-V2.5 is an upgraded version that combines DeepSeek-V2-Chat and DeepSeek-Coder-V2-Instruct, integrating general and coding abilities. It better aligns with human preferences and has been optimized in various aspects, including writing and instruction following.
發布日期
2024-12-10
參數規模
236.0B
上下文長度
164K
支援模態
text
能力雷達圖
13
general
60
coding
76
reasoning
68
science估算
0
agents
0
multimodal
Science 在缺少專門科學評測時使用推理能力代理估算。
排行榜排名
基準測試分數 (LLM Stats)
Code
HumanEval
89.0%自報
Aider
72.2%自報
SWE-Bench Verified
16.8%自報
Communication
MT-Bench
0.90 / 100自報
Creativity
AlignBench
80.4%自報
Arena Hard
76.2%自報
AlpacaEval 2.0
50.5%自報
Finance
MMLU
80.4%自報
General
DS-FIM-Eval
78.3%自報
LiveCodeBench(01-09)
41.8%自報
Language
BBH
84.3%自報
Math
GSM8k
95.1%自報
MATH
74.7%自報
Reasoning
HumanEval-Mul
73.8%自報
DS-Arena-Code
63.1%自報
AA 評測指數
Intelligence Index12.5
Math 5000.8
LLM Stats 分類評分
Communication90
Roleplay90
Finance80
General80
Healthcare80
Language80
Legal80
Math80
Writing70
Creativity70
Reasoning70
Code60
Frontend Development20
定價
輸入價格免費
輸出價格免費
混合價格(3:1)免費
速度
Tokens/秒0.0 tokens/s
首Token延遲0.00s
首回答延遲0.00s
可用提供商
(LS 內部計價單位)暫無提供商資料