AIEvals

Benchmarks

Apex Bench v1.1

The evaluation named Apex Bench v1.1 in DeepSeek’s 17 October 2026 model card.

ReasoningReported by DeepSeekVendor-reported onlyReported by DeepSeek

How it is run

DeepSeek publishes six percentage scores in its dated release comparison, including 72.1 for Flash 1017 and 64.6 for the September Flash model. The source does not describe the task composition, effort, harness, judge or trial count.

What it cannot tell you

The source does not establish equivalence with MathArena Apex, APEX-Agents-AA or any other Apex-named evaluation. This version retains its own page and publisher ranks; its scores are not imported into those other benchmarks.

Why there is no independent table

No independent suite in this dataset runs Apex Bench v1.1. The numbers below are the ones model publishers report about their own models, which is a different kind of evidence: each publisher chooses its own harness and effort level, so a row is readable on its own and not against the other rows.

Read this with

Nothing else on this site shares this benchmark's family or category, so these are other evaluations the same publisher reports.

  • Humanity's Last Exam (text-only, without tools)Expert-level reasoning on the text-only subset of Humanity’s Last Exam, with no image questions or tools.
  • Terminal-Bench 4.0Whether an agent can complete a long, realistic piece of work in a sandboxed terminal: a working service, a proof, a CAD model, a trained kernel, a forensic report.
  • ProgramBenchWhether an agent can produce a program that fully satisfies a specification, scored on complete resolution rather than partial credit.
  • AutomationBenchWhether an agent can complete an end-to-end business workflow across applications — sales, marketing, operations, support, finance and HR — using 47 tools.

As reported by each publisher

Kept separate per publisher, because the settings are not shared.

DeepSeek

DeepSeek-V4.1-Flash-1017: complete release comparison · 17 October 2026

RankModelScore
1DeepSeek V4.1 Flash 101772.1%
3DeepSeek V4.1 Flash64.6%
6Kimi K347.8%
4GLM 5.356.6%
5GPT-5.6 Sol54.3%
2Claude Opus 565.8%

Ranks compare reported scores in the same published row; ties share a rank.

Apex Bench v1.1. Apex Bench v1.1 as named in this model card. This is not the MathArena Apex result of 65.6 reported for the September model, nor APEX-Agents-AA.

The complete 11-row, six-model comparison from DeepSeek’s dated MIT model card, read 17 October 2026 at revision 1bae0e1088993b9b82de08f59de5be5a34e42170. All 63 published comparison cells and both additional parenthesized HLE scores are retained as 65 figures across 12 rows. The HLE text-only scores are split into their own row and benchmark page; GLM-5.3’s dagger-marked 42.0 is text-only and has no full-set score. Unreported values stay null and do not enter ranks. K3 and Opus 5 are displayed as Kimi K3 and Claude Opus 5. Terminal-Bench 4.0 uses Claude Code, DeepSWE v1.1 uses Mini-SWE-V2, and the release notes identify ZeroBench-main as Pass@5. Other effort levels, trial counts and judges are not fully specified. These are DeepSeek publication claims, separate from independent evaluations and earlier publisher snapshots.

huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017

Sources

Every page used for this benchmark, read on 17 October 2026.