DeepSeek-V3.2-Speciale · LiveCodeBench(DeepSeek-V3.2 report Table 3; Speciale explicitly named; LiveCodeBench (Pass@1-COT); unit not printed) 88.7 · 附件证据(deepseek-v3-2-speciale-ev-005)

DeepSeek-V3.2-Speciale 的 LiveCodeBench · DeepSeek-V3.2 report Table 3; Speciale explicitly named; LiveCodeBench (Pass@1-COT); unit not printed 官方出处。DeepSeek-V3.2-Speciale · 88.7。见《DeepSeek-V3.2-Speciale technical report》PDF p.15; Table 3; full header and target row。证据 deepseek-v3-2-speciale-ev-005。

原文与定位

DeepSeek-V3.2-Speciale · LiveCodeBench(DeepSeek-V3.2 report Table 3; Speciale explicitly named; LiveCodeBench (Pass@1-COT); unit not printed),DeepSeek-V3.2-Speciale配置,Pass@1-CoT 88.7(具体成绩公布时间未确认)。见《DeepSeek-V3.2-Speciale technical report》PDF p.15; Table 3; full header and target row。 查看成绩条件

Benchmark | GPT-5 High | Gemini-3.0 Pro | Kimi-K2 Thinking | DeepSeek-V3.2 Thinking | DeepSeek-V3.2 Speciale | LiveCodeBench (Pass@1-COT) | 84.5 (13k) | 90.7 (13k) | 82.6 (29k) | 83.3 (16k) | 88.7 (27k) |
官方材料
DeepSeek-V3.2-Speciale technical report · PDF 第 15 页
定位
附件 · PDF p.15; Table 3; full header and target row
材料版本
Official V3.2 report: 23/23 PDF pages read; Table 3 and Table 4 were checked. Only the explicitly named Speciale columns and rows were attributed to Speciale.
版本公布
未确认
披露时间
未确认 · 首发/后补未确认
采集时间
历史依据
暂无可靠历史版本依据
引用文本

DeepSeek-V3.2-Speciale · LiveCodeBench(DeepSeek-V3.2 report Table 3; Speciale explicitly named; LiveCodeBench (Pass@1-COT); unit not printed),DeepSeek-V3.2-Speciale配置,Pass@1-CoT 88.7(具体成绩公布时间未确认)。官方出处 DeepSeek-V3.2-Speciale technical report(附件,Official V3.2 report: 23/23 PDF pages read; Table 3 and Table 4 were checked. Only the explicitly named Speciale columns and rows were attributed to Speciale.)PDF p.15; Table 3; full header and target row,https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Speciale/resolve/main/assets/paper.pdf?download=true#page=15。Benchlog https://benchlog.lekshome.cc/evidence/v3-2-speciale/deepseek-v3-2-speciale-ev-005/

关联成绩与条件

DeepSeek-V3.2-Speciale · LiveCodeBench 成绩与配置
指标与成绩配置与评测条件证据关系
Pass@1-CoT88.7原文 88.7 · 单位 未标注
公布 未确认
DeepSeek-V3.2-Speciale
设置
High-compute reasoning variant; card says it is designed exclusively for deep reasoning and does not support tool calling.
未知项
The benchmark graph and linked final submission attachments were not fully read.
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
Official DeepSeek-V3.2 report Table 3; cells contain accuracy and output-token count in thousands; only the accuracy value is recorded.
条件
DeepSeek-V3.2 Speciale column; output-token count retained in raw row but not converted into a benchmark score.

直接支持该成绩

在排名中查看此成绩