Benchmark Almanac

MK-Bench

DeepSeek-V3.2 6.83% leads since January 2026
Not in model cards

Agents schedule jobs when machines break down or orders arrive or disappear. The grader compares each completion time with the best known schedule.

The RACE-Sched authors tested six model backends on the ten dynamic Brandimarte instances. The release preserves the original machine choices and processing times.

Made by
Shijie Cao and Yuan Yuan
Released
January 2026
Kind
Tasks from real jobs
Level
University
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Subject
Mathematics
Language
English
In and out
Text → Actions
Interaction
Agentic
Environment
A simulator
Refresh
One fixed set
Run by
Anyone
In views
Agents, Robotics
Strongest model tested
DeepSeek-V3.2 #91 on the capability index
Last model added
December 2025
Scores refreshed
November 20th, 2026
9
Entries in its paper, 5 models we track

DeepSeek-V3.2 leads with 6.83% among models released by December 2025 DeepSeek-V3.2 leads with 6.83% among models released by December 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in December 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 DeepSeek-V3.2 6.83%
  2. 2 Qwen3-8B 10.3%
  3. 3 IDDQN 11.8%
  4. 4 GPT-5 nano 13.2%
  5. 5 Qwen3-32B 13.8%
  6. 6 Qwen3-14B 14.3%
  7. 7 HMPSAC 14.3%
  8. 8 PPO-OC 15.9%
  9. 9 DAN 17.8%

Source: The MK-Bench paper

Source: The MK-Bench paper

#EntryScoreSettingDateSource
1 DeepSeek-V3.2 6.83% ReflecSched Dec 2025 Official leaderboard
2 Qwen3-8B 10.3% ReflecSched Apr 2025 Official leaderboard
3 IDDQN 11.8% Official leaderboard
4 GPT-5 nano 13.2% ReflecSched Aug 2025 Official leaderboard
5 Qwen3-32B 13.8% ReflecSched Apr 2025 Official leaderboard
6 Qwen3-14B 14.3% ReflecSched Apr 2025 Official leaderboard
7 HMPSAC 14.3% Official leaderboard
8 PPO-OC 15.9% Official leaderboard
9 DAN 17.8% Official leaderboard
#ModelScoreSettingDateSource
1 DeepSeek-V3.2 6.83% ReflecSched Dec 2025 Official leaderboard
2 Qwen3-8B 10.3% ReflecSched Apr 2025 Official leaderboard
3 GPT-5 nano 13.2% ReflecSched Aug 2025 Official leaderboard
4 Qwen3-32B 13.8% ReflecSched Apr 2025 Official leaderboard
5 Qwen3-14B 14.3% ReflecSched Apr 2025 Official leaderboard

Rankings by part

We read 10 parts from the benchmark’s paper, each ranking models on one part of the test, such as MK01, MK02 and MK03. DeepSeek-V3.2 leads 4 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

MK01

5 models we track have a score here.

  1. 1 Qwen3-8B ↑1 5.05%
  2. 2 Qwen3-32B ↑2 7.32%
  3. 3 GPT-5 nano 7.39%
  4. 4 Qwen3-14B ↑1 10.7%
  5. 5 DeepSeek-V3.2 ↓4 15.9%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

DeepSeek
Benchmark paper Not tested since Jan 2026

Source: The MK-Bench paper

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2025 2026 DeepSeek-V3.2 Qwen3-8B 0% 20% 40% 60% 80% 100% 2025 2026 DeepSeek-V3.2 Qwen3-32B

Source: Official MK-Bench leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Economic work

All 210 in Economic work
  • JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
  • GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate. Seed2.1 Pro 87.9%
  • GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
  • AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
  • Agents' Last Exam v1.1 In use Agents complete 122 public long-horizon professional workflow tasks across 55 sub-industries. Deterministic graders check outcomes against hidden references. GPT-6.2 Astra 47.5%
  • CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 11 months ago, in December 2025, for DeepSeek-V3.2. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (9 entries).
  • Pass Models scored 5 models we track, plus 4 other entries on its leaderboard.
  • Warning Newest score The newest came 11 months ago, in December 2025, for DeepSeek-V3.2.
  • Note Frontier models None of the top 10 has a score. The strongest tested is DeepSeek-V3.2 (#91).
  • Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 5 rows for models we track and 4 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check