Benchmark Almanac

ExpertLongBench

Gemini 2.5 Pro Jun 2025 33.4% leads since June 2025
Not in model cards

Models write documents for 11 professional tasks. A judge compares facts in each document with a reference checklist and scores precision and recall.

The authors rank 13 models by average F1 across 11 tasks. They release seven tasks and keep four private.

Released
June 2025
Kind
Tasks from real jobs
Level
Expert
Scored
Graded by a model
Answers
Open-ended
Tasks
Public, plus a private test set
Measures
Capability
Language
English
Interaction
Single turn
Refresh
One fixed set
Run by
A host
Occupations
Legal, Educational instruction and library, Healthcare practitioners and technical
Strongest model tested
GPT-5 #71 on the capability index
Last model added
August 2025
Scores refreshed
November 20th, 2026
13
Entries on its own leaderboard, 10 models we track

Gemini 2.5 Pro leads with 33.4% among models released by August 2025 Gemini 2.5 Pro leads with 33.4% among models released by August 2025

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. 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 August 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 Gemini 2.5 Pro Jun 2025 33.4%
  2. 2 GPT-5 31%
  3. 3 Gemini 2.0 Flash Feb 2025 26.8%
  4. 4 GPT-4o Mar 2025 26.5%
  5. 5 GPT-4o mini 26.1%
  6. 6 Llama 3.3 70B Instruct 24.6%
  7. 7 Mistral-Large-Instruct 24%
  8. 8 Mistral-Nemo-Instruct 23.8%
  9. 9 Qwen2.5-72B 23.8%
  10. 10 Llama3.1-8B-Instruct 23.4%
  11. 11 Claude 3.7 Sonnet 23.2%
  12. 12 Qwen2.5-7B 20.5%
  13. 13 Claude 3.5 Haiku 19.3%

Source: Official ExpertLongBench leaderboard

#EntryScoreSettingDateSource
1 Gemini 2.5 Pro Jun 2025 33.4% Jun 2025 Official leaderboard
2 GPT-5 31% Aug 2025 Official leaderboard
3 Gemini 2.0 Flash Feb 2025 26.8% Feb 2025 Official leaderboard
4 GPT-4o Mar 2025 26.5% Mar 2025 Official leaderboard
5 GPT-4o mini 26.1% Jul 2024 Official leaderboard
6 Llama 3.3 70B Instruct 24.6% Dec 2024 Official leaderboard
7 Mistral-Large-Instruct 24% Official leaderboard
8 Mistral-Nemo-Instruct 23.8% Official leaderboard
9 Qwen2.5-72B 23.8% Sep 2024 Official leaderboard
10 Llama3.1-8B-Instruct 23.4% Official leaderboard
11 Claude 3.7 Sonnet 23.2% Feb 2025 Official leaderboard
12 Qwen2.5-7B 20.5% Sep 2024 Official leaderboard
13 Claude 3.5 Haiku 19.3% Oct 2024 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 2.5 Pro Jun 2025 33.4% Jun 2025 Official leaderboard
2 GPT-5 31% Aug 2025 Official leaderboard
3 Gemini 2.0 Flash Feb 2025 26.8% Feb 2025 Official leaderboard
4 GPT-4o Mar 2025 26.5% Mar 2025 Official leaderboard
5 GPT-4o mini 26.1% Jul 2024 Official leaderboard
6 Llama 3.3 70B Instruct 24.6% Dec 2024 Official leaderboard
7 Qwen2.5-72B 23.8% Sep 2024 Official leaderboard
8 Claude 3.7 Sonnet 23.2% Feb 2025 Official leaderboard
9 Qwen2.5-7B 20.5% Sep 2024 Official leaderboard
10 Claude 3.5 Haiku 19.3% Oct 2024 Official leaderboard

Rankings by part

We read 11 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as T1, T2 and T3. GPT-5 leads 5 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

T1

10 models we track have a score here.

  1. 1 GPT-5 ↑1 27.2%
  2. 2 Gemini 2.5 Pro Jun 2025 ↓1 25.4%
  3. 3 GPT-4o mini ↑2 16.5%
  4. 4 GPT-4o Mar 2025 13.2%
  5. 5 Gemini 2.0 Flash Feb 2025 ↓2 13.1%
  6. 6 Qwen2.5-72B ↑1 12.3%
  7. 7 Llama 3.3 70B Instruct ↓1 12.1%
  8. 8 Claude 3.7 Sonnet 11.5%
  9. 9 Qwen2.5-7B 10.7%
  10. 10 Claude 3.5 Haiku 2.80%

2 models have held number one on the official leaderboard since its release in June 2025

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

Google DeepMind
Own leaderboard Not tested since Aug 2025

Source: Official ExpertLongBench 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 Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Gemini 2.5 Pro Gemini 2.0 Flash GPT-4o mini 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Gemini 2.5 Pro Gemini 2.0 Flash

Source: Official ExpertLongBench 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

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 15 months ago, in August 2025, for GPT-5. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (13 entries).
  • Pass Models scored 10 models we track, plus 3 other entries on its leaderboard.
  • Warning Newest score The newest came 15 months ago, in August 2025, for GPT-5.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
  • Pass Headroom The best is 33.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 10 rows for models we track and 3 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