DeepSeek-V3.1-Terminus · 官方评测披露

DeepSeek-V3.1-Terminus 当前收录 12 项测试的官方记录。已完成列明材料的核查。

模型发布 · DeepSeek V3

官方材料与核查范围

All listed official model-card, official News/API, metadata and directly linked technical-report materials for this release were read through the requested scope. Tables without an explicit unit retain an empty unit; figures without exact labels were not estimated. 核对时间 。

当前资料不足以还原首发。 No reliable historical snapshot proving the complete launch-day material set was obtained. Official publication dates and current materials do not substitute for a launch snapshot.

  • DeepSeek-V3.1-Terminus official Hugging Face model card · 主文 · HF revision 19510d6dc61f79dbd925bd51ee8a9081c509a4b6; current card snapshot
    公布 未确认 · 采集 · 已核查 Official model-card Markdown read in full; benchmark rows, notes, configuration statements and linked reports checked.
  • DeepSeek-V3.1-Terminus official Hugging Face model metadata · 主文 · author=deepseek-ai; repo createdAt=2025-09-22T04:56:04.000Z (repository upload time only); lastModified=2025-09-29T01:52:07.000Z; sha=19510d6dc61f79dbd925bd51ee8a9081c509a4b6
    公布 未确认 · 采集 · 已核查 Official model API metadata checked for author, repository identity, revision and timestamps; createdAt was never used as release date.
  • DeepSeek-V3.1 版本更新 · 主文 · Official DeepSeek News index entry dated 2025-09-22; index downloaded and retained locally
    公布 · 采集 · 已核查 The official dated entry or current service table was read for this release scope; aliases were not treated as weight checkpoints.

配置

  • DeepSeek-V3.1-Terminus reported comparison · The table does not specify an effort setting; search-agent template/tool-set changed as stated in the card.
    未知项 Do not infer a reasoning mode from the predecessor comparison.

直接前代与后续发布

直接前代 DeepSeek-V3.1。 官方关系依据

DeepSeek-V3.1 更新至 Terminus 版本,缓解中英文混杂问题,优化 Code Agent 与 Search Agent 表现。
9 项披露延续 · 3 项新增 · 4 项本次未见 · 0 项待核查
DeepSeek-V3.1 与 DeepSeek-V3.1-Terminus 的披露变化
测试已核查状态DeepSeek-V3.1DeepSeek-V3.1-Terminus
MMLU-Pro · DeepSeek card version unspecified; Exact Match延续

已记录的评测条件有变化

前代记录本次记录
GPQA Diamond · Diamond subset; task snapshot unspecified延续

已记录的评测条件有变化

前代记录本次记录
Humanity's Last Exam · Text-only subset as stated in V3.1 and later comparison cards; task revision unspecified延续

已记录的评测条件有变化

前代记录本次记录
LiveCodeBench · Version unspecified; Pass@1新增—本次记录
Codeforces · Rating; contest snapshot unspecified新增—本次记录
Aider-Polyglot · Version unspecified; Accuracy新增—本次记录
BrowseComp · Version unspecified; Accuracy延续

已记录的评测条件有变化

前代记录本次记录
BrowseComp-zh · Version unspecified; Accuracy延续

已记录的评测条件有变化

前代记录本次记录
SimpleQA · Internal search agent protocol; task version unspecified延续

已记录的评测条件有变化

前代记录本次记录
SWE-bench Verified · Agent mode; task snapshot unspecified延续

已记录的评测条件有变化

前代记录本次记录
SWE-bench Multilingual · Agent mode; task snapshot unspecified延续

已记录的评测条件有变化

前代记录本次记录
Terminal-Bench · Terminus 1 framework; task version unspecified延续

已记录的评测条件有变化

前代记录本次记录
LiveCodeBench · 2408-2505; Pass@1本次未见前代记录—
Codeforces-Div1 · Div1; contest snapshot unspecified本次未见前代记录—
AIME · 2025 edition; Pass@1本次未见前代记录—
HMMT · 2025 competition; Pass@1本次未见前代记录—

测试记录 · 12