DeepSeek-V4.1-Flash-1017 · 官方评测披露

DeepSeek-V4.1-Flash-1017 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 V4.1 Flash

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-V4.1-Flash-1017 official Hugging Face model card · 主文 · HF revision 08c8657f02291e11b6f6b63e0e16014f8e748518; 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-V4.1-Flash-1017 official Hugging Face model metadata · 主文 · author=deepseek-ai; repo createdAt=2026-10-17T02:40:38.000Z (repository upload time only); lastModified=2026-11-10T02:22:56.000Z; sha=08c8657f02291e11b6f6b63e0e16014f8e748518
    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 API Lists Models · 主文 · Current official API model list snapshot downloaded 2026-11-17; alias deepseek-flash note identifies V4.1-Flash-1017 as serving model
    公布 未确认 · 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.
  • DeepSeek-V4.1-Flash Technical Report · 附件 · Official V4.1-Flash report: 51/51 PDF pages read; Table 1, Table 3, Table 4, Table 5, multi-agent results and exact-label results were checked. CyberGym, SEC-Bench Pro and ExploitGym were retained as local audit only and excluded from regular mappings. No generic V4.1-Flash report value was copied into the 1017 checkpoint.
    公布 未确认 · 采集 · 已核查 Official V4.1-Flash report: 51/51 PDF pages read; Table 1, Table 3, Table 4, Table 5, multi-agent results and exact-label results were checked. CyberGym, SEC-Bench Pro and ExploitGym were retained as local audit only and excluded from regular mappings. No generic V4.1-Flash report value was copied into the 1017 checkpoint.
  • DeepSeek API Change Log · 主文 · 当前官方更新日志;定位:## Date: 2026-10-17 / ### DeepSeek-V4.1-Flash-1017 Update。只核读该日期与相关发布段落;原始 Markdown 已留存。
    公布 未确认 · 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-V4.1-Flash-1017 reported comparison · The table reports this named checkpoint; reasoning_effort and most evaluation settings are not stated in this card.
    未知项 Do not copy reasoning_effort=100 from the separate V4.1-Flash card.、Do not merge these scores with DeepSeek-V4.1-Flash.

直接前代与后续发布

直接前代 DeepSeek-V4.1-Flash。 官方关系依据

**DeepSeek-V4.1-Flash-1017** is a minor update to [**DeepSeek-V4.1-Flash**](https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash), delivering substantial improvements across reasoning, coding, agentic tasks, and visual benchmarks.
0 项披露延续 · 12 项新增 · 22 项本次未见 · 0 项待核查
DeepSeek-V4.1-Flash 与 DeepSeek-V4.1-Flash-1017 的披露变化
测试已核查状态DeepSeek-V4.1-FlashDeepSeek-V4.1-Flash-1017
Humanity's Last Exam · V4.1-Flash-1017 comparison table; non-text-only value, metric/set unspecified新增—本次记录
Terminal-Bench · 4.0; metric not printed in V4.1-Flash-1017 card新增—本次记录
DeepSWE · v1.1; metric not printed in V4.1-Flash-1017 card新增—本次记录
ProgramBench · V4.1-Flash-1017 card does not print metric or task version新增—本次记录
NL2Repo-Bench · Metric/task version unspecified in 1017 card新增—本次记录
AutomationBench · V1.0.6 label; metric not stated in V4.1-Flash-1017 card新增—本次记录
Agents' Last Exam · Metric and task revision unspecified in V4.1-Flash-1017 card新增—本次记录
Apex Bench · v1.1; Pass@1 as printed in V4.1-Flash-1017 card新增—本次记录
Chartography · With tools; metric/task version unspecified in 1017 card新增—本次记录
BabyVision · With tools; metric/task version unspecified in 1017 card新增—本次记录
ZeroBench-main · With tools; metric not printed in 1017 card新增—本次记录
Humanity's Last Exam · V4.1-Flash-1017 comparison table; text-only † value, metric/set unspecified新增—本次记录
GPQA Diamond · Diamond subset; task snapshot unspecified本次未见前代记录—
Humanity's Last Exam · Full HLE set as distinguished from the card-marked text-only subset; task revision unspecified本次未见前代记录—
Codeforces · Rating; contest snapshot unspecified本次未见前代记录—
MathArena Apex · Version unspecified; Pass@1 as printed in V4.1-Flash card本次未见前代记录—
Terminal-Bench · 2.1; Pass@1 as printed in V4.1-Flash card本次未见前代记录—
Terminal-Bench · 3.0; Pass@1 as printed in V4.1-Flash card本次未见前代记录—
Terminal-Bench · 4.0; Pass@1 as printed in V4.1-Flash card本次未见前代记录—
DeepSWE · v1.1; Resolved metric as printed in V4.1-Flash card本次未见前代记录—
ProgramBench · Metric Almost@1 as printed in V4.1-Flash card; task version unspecified本次未见前代记录—
NL2Repo-Bench · Version unspecified; Score metric as printed in V4.1-Flash card本次未见前代记录—
AutomationBench · v1.0.6; Pass@1 as printed in V4.1-Flash card本次未见前代记录—
Agents' Last Exam · Task revision unspecified; Pass@1 as printed in V4.1-Flash card本次未见前代记录—
Humanity's Last Exam · With-tools evaluation; task subset unspecified本次未见前代记录—
Chartography · With tools; task revision unspecified本次未见前代记录—
BabyVision · With tools; task revision unspecified本次未见前代记录—
ZeroBench-main · With tools; Pass@5 as printed in V4.1-Flash card本次未见前代记录—
Humanity's Last Exam · Text-only subset marked † in DeepSeek V4.1-Flash card; task revision unspecified本次未见前代记录—
Terminal-Bench · 2.1 (Pass@1); across agent scaffolds at Max reasoning effort; unit not printed本次未见前代记录—
ProgramBench · High-confidence 172-task subset; Almost@1; multi-agent; 8-hour deadline本次未见前代记录—
ProgramBench · High-confidence 172-task subset; Almost@1; single-agent; 8-hour deadline本次未见前代记录—
FrontierSWE v2 · No-GPU public subset; Mean@5; multi-agent; 20-hour deadline本次未见前代记录—
FrontierSWE v2 · No-GPU public subset; Mean@5; single-agent; 20-hour deadline本次未见前代记录—

测试记录 · 12