GDPval
Seed2.1 Pro 87.9% leads since Jun 2026 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.
Only the 220-task gold set is public, with an automated grading service. OpenAI says its hosted GDPval leaderboard is no longer active and points to the GDPval-AA board instead.
In the authors’ words
We introduce GDPval, a benchmark evaluating AI model capabilities on real-world economically valuable tasks. GDPval covers the majority of U.S. Bureau of Labor Statistics Work Activities for 44 occupations across the top 9 sectors contributing to U.S. GDP (Gross Domestic Product). Tasks are constructed from the representative work of industry professionals with an average of 14 years of experience. We find that frontier model performance on GDPval is improving roughly linearly over time, and that the current best frontier models are approaching industry experts in deliverable quality. We analyze the potential for frontier models, when paired with human oversight, to perform GDPval tasks cheaper and faster than unaided experts. We also demonstrate that increased reasoning effort, increased task context, and increased scaffolding improves model performance on GDPval. Finally, we open-source a gold subset of 220 tasks and provide a public automated grading service at evals.openai.com to facilitate future research in understanding real-world model capabilities.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Graded by people Human experts grade or compare the outputs.
- Answers
- Open-ended Free-form text, documents, designs or media judged as a whole.
Seed2.1 Pro leads with 87.9% among models released by Jun 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Jun 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 13 more
- 1 Seed2.1 Pro 87.9%
- 2 GPT-5.5 84.9%
- 3 GPT-5.4 83%
- 4 GPT-5.5 Pro 82.3%
- 5 GPT-5.4 Pro 82%
- 6 GPT-5.2 Pro 74.1%
- 7 GPT-5.2 70.9%
- 8 GPT-5.3 Codex 70.9%
- 9 Seed2.0 Lite (0428) 53.1%
- 10
Nemotron 3 Ultra 46.7%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Seed2.1 Pro | 87.9% | Jun 2026 | Model card | |
| 2 | GPT-5.5 | 84.9% | Apr 2026 | Model card | |
| 3 | GPT-5.4 | 83% | Mar 2026 | Model card | |
| 4 | GPT-5.5 Pro | 82.3% | Apr 2026 | Model card | |
| 5 | GPT-5.4 Pro | 82% | Mar 2026 | Model card | |
| 6 | GPT-5.2 Pro | 74.1% | Dec 2025 | Model card | |
| 7 | GPT-5.2 | 70.9% | Dec 2025 | Model card | |
| 8 | GPT-5.3 Codex | 70.9% | Feb 2026 | Model card | |
| 9 | Seed2.0 Lite (0428) | 53.1% | Apr 2026 | Model card | |
| 10 | Nemotron 3 Ultra | 46.7% | Jun 2026 | Model card |
Epoch AI has scores for 11 models. We plot every source in the charts below.
3 models have held number one since its release in Sep 2025
Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for Epoch AI’s runs and what labs claim in their cards.
The best score passed 85% in Jun 2026
All the scores we have, one dot each, so a model tested at several settings appears more than once.
3 labs put it in a launch post in the last six months
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 21 Sep 2026 | Grok 4.7 | 1695 | GDPval (Professional knowledge work) Elo chart [Grok 4.7 xhigh] |
| 23 Jun 2026 | Seed2.1 Pro | 87.9 | GDPval (model card Table 1) |
| 4 Jun 2026 | Nemotron 3 Ultra | 46.7 | GDPVal (metric not stated) |
| 28 Apr 2026 | Seed2.0 Lite (0428) | 53.1 | GDPval |
| 23 Apr 2026 | GPT-5.5 Pro | 82.3 | GDPval (wins or ties) |
| 23 Apr 2026 | GPT-5.5 | 84.9 | GDPval (wins or ties) |
| 5 Mar 2026 | GPT-5.4 Pro | 82 | GDPval |
| 5 Mar 2026 | GPT-5.4 | 83 | GDPval (wins or ties) |
| 5 Feb 2026 | GPT-5.3 Codex | 70.9 | GDPval (wins or ties) |
| 11 Dec 2025 | GPT-5.2 Pro | 74.1 | GDPval (ties allowed, wins or ties) |
| 11 Dec 2025 | GPT-5.2 | 70.9 | GDPval (wins or ties) Knowledge work tasks |
2 system cards report it, from 2 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 23 Jun 2026 | Seed2.1 Pro model card | 87.9 | GDPval (Table 1) | Capabilities |
| 9 Jun 2026 | Nemotron 3 Ultra technical report | 46.7 | GDPVal, NVIDIA open implementation, metric not stated (Table 10) | Capabilities |
More 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-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
- CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished).
Qwen3.8-Omni-Flash 75.3% - OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%
- SpreadsheetBench In use Real spreadsheet manipulation questions from Excel forums. Output workbook is checked cell-by-cell against reference solutions.
MiniMax-M3 89.3%