Benchmark Almanac

AEC-Bench (Theodore Galanos)

Grok 4.3 0.89 leads since May 2026
Not in model cards

Agents perform engineering calculations such as cable sizing and hydraulic design. Task verifiers score their outputs, and the leaderboard reports mean reward and completion.

The public leaderboard covers 552 tasks and reports completion alongside mean reward. Galanos also coauthored Nomic’s separate construction document benchmark.

Released
May 2026
Kind
Tasks from real jobs
Level
University
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions, Text
Interaction
Agentic
Environment
Terminal
Items from
Generated
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
GPT-5.2 #51 on the capability index
Last model added
April 2026
Scores refreshed
November 20th, 2026
18
Entries on its own leaderboard, 17 models we track

Grok 4.3 leads with 0.89 among models released by April 2026 Grok 4.3 leads with 0.89 among models released by April 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. 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 April 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Grok 4.3 0.89
  2. 2 Grok 4.20 reasoning 0.87
  3. 3 Kimi K2.6 0.86
  4. 4 GPT-5.2 0.83
  5. 5 DeepSeek-V4-Flash 0.83
  6. 6 GPT-5.3 Instant 0.79
  7. 7 MiniMax-M2.7 0.73
  8. 8 Grok 4.1 Fast reasoning 0.73
  9. 9 Mistral Large 3 0.71
  10. 10 GPT-5.1 0.70
  11. 11 DeepSeek-V4-Pro 0.68
  12. 12 Qwen3.5-9B 0.58
  13. 13 DeepSeek-V3.2-Speciale 0.51
  14. 14 gpt-oss-120b 0.50
  15. 15 Gemma 4 31B IT 0.47
  16. 16 GPT-4.1 mini 0.36
  17. 17 LFM2 24B A2B 0.33
  18. 18 Ministral 3B 0.21

Source: Official AEC-Bench (Theodore Galanos) leaderboard

#EntryScoreSettingReleasedRun onSource
1 Grok 4.3 0.89 tool_loop Apr 2026 May 2026 Official leaderboard
2 Grok 4.20 0.87 reasoning, tool_loop Feb 2026 May 2026 Official leaderboard
3 Kimi K2.6 0.86 tool_loop Apr 2026 May 2026 Official leaderboard
4 GPT-5.2 0.83 tool_loop Dec 2025 May 2026 Official leaderboard
5 DeepSeek-V4-Flash 0.83 tool_loop Apr 2026 May 2026 Official leaderboard
6 GPT-5.3 Instant 0.79 tool_loop Mar 2026 May 2026 Official leaderboard
7 MiniMax-M2.7 0.73 tool_loop Mar 2026 May 2026 Official leaderboard
8 Grok 4.1 Fast 0.73 reasoning, tool_loop Nov 2025 May 2026 Official leaderboard
9 Mistral Large 3 0.71 tool_loop Dec 2025 May 2026 Official leaderboard
10 GPT-5.1 0.70 tool_loop Nov 2025 May 2026 Official leaderboard
11 DeepSeek-V4-Pro 0.68 tool_loop Apr 2026 May 2026 Official leaderboard
12 Qwen3.5-9B 0.58 tool_loop Feb 2026 May 2026 Official leaderboard
13 DeepSeek-V3.2-Speciale 0.51 tool_loop Dec 2025 May 2026 Official leaderboard
14 gpt-oss-120b 0.50 tool_loop Aug 2025 May 2026 Official leaderboard
15 Gemma 4 31B IT 0.47 tool_loop Apr 2026 May 2026 Official leaderboard
16 GPT-4.1 mini 0.36 tool_loop Apr 2025 May 2026 Official leaderboard
17 LFM2 24B A2B 0.33 tool_loop May 2026 Official leaderboard
18 Ministral 3B 0.21 tool_loop Oct 2024 May 2026 Official leaderboard
#ModelScoreSettingReleasedRun onSource
1 Grok 4.3 0.89 tool_loop Apr 2026 May 2026 Official leaderboard
2 Grok 4.20 0.87 reasoning, tool_loop Feb 2026 May 2026 Official leaderboard
3 Kimi K2.6 0.86 tool_loop Apr 2026 May 2026 Official leaderboard
4 GPT-5.2 0.83 tool_loop Dec 2025 May 2026 Official leaderboard
5 DeepSeek-V4-Flash 0.83 tool_loop Apr 2026 May 2026 Official leaderboard
6 GPT-5.3 Instant 0.79 tool_loop Mar 2026 May 2026 Official leaderboard
7 MiniMax-M2.7 0.73 tool_loop Mar 2026 May 2026 Official leaderboard
8 Grok 4.1 Fast 0.73 reasoning, tool_loop Nov 2025 May 2026 Official leaderboard
9 Mistral Large 3 0.71 tool_loop Dec 2025 May 2026 Official leaderboard
10 GPT-5.1 0.70 tool_loop Nov 2025 May 2026 Official leaderboard
11 DeepSeek-V4-Pro 0.68 tool_loop Apr 2026 May 2026 Official leaderboard
12 Qwen3.5-9B 0.58 tool_loop Feb 2026 May 2026 Official leaderboard
13 DeepSeek-V3.2-Speciale 0.51 tool_loop Dec 2025 May 2026 Official leaderboard
14 gpt-oss-120b 0.50 tool_loop Aug 2025 May 2026 Official leaderboard
15 Gemma 4 31B IT 0.47 tool_loop Apr 2026 May 2026 Official leaderboard
16 GPT-4.1 mini 0.36 tool_loop Apr 2025 May 2026 Official leaderboard
17 Ministral 3B 0.21 tool_loop Oct 2024 May 2026 Official leaderboard

Rankings by part

We read 5 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Civil, Electrical and Ground. Kimi K2.6 leads 3 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Civil

17 models we track have a score here.

  1. 1 Kimi K2.6 ↑2 0.88
  2. 2 Grok 4.20 0.86
  3. 3 Grok 4.3 ↓2 0.85
  4. 4 DeepSeek-V4-Flash ↑1 0.82
  5. 5 DeepSeek-V4-Pro ↑6 0.80
  6. 6 GPT-5.2 ↓2 0.78
  7. 7 GPT-5.3 Instant ↓1 0.78
  8. 8 GPT-5.1 ↑2 0.76
  9. 9 gpt-oss-120b ↑5 0.76
  10. 10 Grok 4.1 Fast ↓2 0.74

The best score, 0.89, costs $0.03 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Mistral AI OpenAI xAI Qwen MiniMax DeepSeek Cost frontier One model at each effort level
0.00 0.20 0.40 0.60 0.80 1.00 $0.0001 $0.001 $0.01 $0.10 Cost per task, log scale Grok 4.3 DeepSeek-V4-Flash MiniMax-M2.7 Qwen3.5-9B gpt-oss-120b Ministral 3B 0.00 0.20 0.40 0.60 0.80 1.00 $0.0001 $0.001 $0.01 $0.10 Cost per task, log scale Grok 4.3 DeepSeek-V4-Flash

Source: Official AEC-Bench (Theodore Galanos) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Grok 4.3 tool_loop 0.89 $0.03 929
Grok 4.20 reasoning tool_loop 0.87 $0.03 952
Kimi K2.6 tool_loop 0.86 $0.07 6,560
GPT-5.2 tool_loop 0.83 $0.03 1,299
DeepSeek-V4-Flash tool_loop 0.83 $0.008 3,029
GPT-5.3 Instant tool_loop 0.79 $0.02 977
MiniMax-M2.7 tool_loop 0.73 $0.007 3,037
Grok 4.1 Fast reasoning tool_loop 0.73 $0.005 1,075
Mistral Large 3 tool_loop 0.71 $0.006 2,049
GPT-5.1 tool_loop 0.70 $0.03 2,264
DeepSeek-V4-Pro tool_loop 0.68 13,578
Qwen3.5-9B tool_loop 0.58 $0.003 3,594
DeepSeek-V3.2-Speciale tool_loop 0.51 14,587
gpt-oss-120b tool_loop 0.50 $0.002 2,311
Gemma 4 31B IT tool_loop 0.47 4,340
GPT-4.1 mini tool_loop 0.36 $0.007 1,934
Ministral 3B tool_loop 0.21 $0.0001 266

The best score, 0.89, costs $0.03 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Mistral AI OpenAI xAI Qwen MiniMax DeepSeek Cost frontier One model at each effort level
0.00 0.20 0.40 0.60 0.80 1.00 $0.0001 $0.001 $0.01 $0.10 Cost per task, log scale Grok 4.3 DeepSeek-V4-Flash MiniMax-M2.7 Qwen3.5-9B gpt-oss-120b Ministral 3B 0.00 0.20 0.40 0.60 0.80 1.00 $0.0001 $0.001 $0.01 $0.10 Cost per task, log scale Grok 4.3 DeepSeek-V4-Flash

Source: Official AEC-Bench (Theodore Galanos) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Grok 4.3 tool_loop 0.89 $0.03 929
Grok 4.20 reasoning tool_loop 0.87 $0.03 952
Kimi K2.6 tool_loop 0.86 $0.07 6,560
GPT-5.2 tool_loop 0.83 $0.03 1,299
DeepSeek-V4-Flash tool_loop 0.83 $0.008 3,029
GPT-5.3 Instant tool_loop 0.79 $0.02 977
MiniMax-M2.7 tool_loop 0.73 $0.007 3,037
Grok 4.1 Fast reasoning tool_loop 0.73 $0.005 1,075
Mistral Large 3 tool_loop 0.71 $0.006 2,049
GPT-5.1 tool_loop 0.70 $0.03 2,264
DeepSeek-V4-Pro tool_loop 0.68 13,578
Qwen3.5-9B tool_loop 0.58 $0.003 3,594
DeepSeek-V3.2-Speciale tool_loop 0.51 14,587
gpt-oss-120b tool_loop 0.50 $0.002 2,311
Gemma 4 31B IT tool_loop 0.47 4,340
GPT-4.1 mini tool_loop 0.36 $0.007 1,934
Ministral 3B tool_loop 0.21 $0.0001 266

Number one over time

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

xAI
Own leaderboard Not tested since May 2026
Grok 4.3 0.89

Source: Official AEC-Bench (Theodore Galanos) leaderboard

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
On the benchmark’s own leaderboard
0.40 0.60 0.80 1.00 2025 2026 0.40 0.60 0.80 1.00 2025 2026

Source: Official AEC-Bench (Theodore Galanos) 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 7 months ago, in April 2026, for Grok 4.3. 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 official leaderboard (18 entries).
  • Pass Models scored 17 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 7 months ago, in April 2026, for Grok 4.3.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on aecbench.com: 17 rows for models we track and 1 other entry 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