GDPval
Win rate against industry professionals on economically valuable tasks (OpenAI).
As of 25 Oct 2026, GPT-5.2 leads GDPval on BenchLeader with 49.7%, ahead of Claude Opus 4.5 at 45.5%, across 11 model configurations with a published result.
- Published by
- OpenAIdata via Epoch AI Benchmarking Hub
- Category
- Agents & tools
- Index weight
- 0.5
- Models
- 11
- Data as of
- 25 Oct 2026
- Newest result
- 11 Dec 2025
The publisher has posted no new results on this board since 11 Dec 2025. BenchLeader re-reads it every day, so newer models appear here as soon as they are published.
CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.
What the test looks like
GDPval asks models to produce real work products, such as a legal memo, a financial model or a design brief, across 44 occupations that carry the most economic weight. Industry professionals then compare each output against a human expert's.
How it is scored
Win rate: how often graders preferred (or tied) the model's deliverable against the professional's. Published by OpenAI.
What to keep in mind
Grading is human and pairwise, so the score is relative and expensive to reproduce; only a gold subset is public. The tasks are one-shot deliverables, not the iterative work most jobs involve.
- 1GPT-5.2 (none)49.7%
- 2Claude Opus 4.545.5%
- 3Claude Opus 4.143.6%
- 4Claude Sonnet 4.542.5%
- 5Gemini 3 Pro40.3%
- 6GPT-5 (medium)34.8%
- 7o3 (medium)30.8%
- 8o4-mini (high)25.3%
- 9Gemini 2.5 Pro23.3%
- 10Grok 4 (high)21.1%
- 11GPT-4o9.9%
| # | ||||
|---|---|---|---|---|
| 1 |  GPT-5.2no reasoning | 49.7% | 48.8 | 2025-12-11 |
| 2 |  Claude Opus 4.5 | 45.5% | 57.9 | 2025-11-24 |
| 3 | Claude Opus 4.1 | 43.6% | 51.5 | 2025-08-05 |
| 4 | Claude Sonnet 4.5 | 42.5% | 53.7 | 2025-09-29 |
| 5 |  Gemini 3 Pro | 40.3% | 60.4 | 2025-11-18 |
| 6 | GPT-5medium | 34.8% | 57.5 | 2025-08-07 |
| 7 | o3medium | 30.8% | 53.3 | 2025-04-16 |
| 8 | o4-minihigh | 25.3% | 50.1 | 2025-04-16 |
| 9 | Gemini 2.5 Pro | 23.3% | 53.9 | 2025-06-17 |
| 10 |  Grok 4high | 21.1% | – | 2025-07-09 |
| 11 | GPT-4o | 9.9% | 43.9 | 2024-11-20 |
More agents & tools benchmarks · All LLM benchmarks
Cite as: BenchLeader, “GDPval leaderboard”, data as of 25 Oct 2026, https://www.benchleader.com/benchmarks/gdpval.
GDPval: questions
- What does GDPval measure?
- GDPval asks models to produce real work products, such as a legal memo, a financial model or a design brief, across 44 occupations that carry the most economic weight. Industry professionals then compare each output against a human expert's. Scores are reported in percent of tasks solved; higher is better.
- Which AI model leads GDPval?
- GPT-5.2 leads GDPval with 49.7% as of 25 Oct 2026, ahead of Claude Opus 4.5 at 45.5%.
- How many models have GDPval results?
- 11 model configurations have a GDPval result on BenchLeader, all taken from OpenAI via Epoch AI Benchmarking Hub.
- Who runs GDPval and how often is it updated?
- GDPval is published by OpenAI. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does GDPval count toward the BenchLeader Index?
- Yes. GDPval contributes to the agents & tools category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.