Best LLM for professional work (GDPval)
- GPT-5.2 leads
- 49.7%
- models ranked
- 9
- lead over #2
- +4.2 pts
GDPval · read October 21
As of October 21, 2026, OpenAI GPT-5.2 leads 9 models on GDPval with 49.7%, ahead of Anthropic Claude Opus 4.5 (45.5%) and Anthropic Claude Opus 4.1 (43.6%).
This list narrows the best LLM for agents ranking.
| Rank | Model | GDPval | Coding evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | GPT-5.2 OpenAI | 49.7% | 58.9 | $14.00 | |
| 2 | Claude Opus 4.5 Anthropic | 45.5% | 66.8 | $25.00 | |
| 3 | Claude Opus 4.1 Anthropic | 43.6% | 27.2 | $75.00 | |
| 4 | Claude Sonnet 4.5 Anthropic | 42.5% | 59.1 | $15.00 | |
| 5 | GPT-5 OpenAI | 34.8% | 72.2 | $10.00 | |
| 6 | o3 OpenAI | 30.8% | 70.7 | $8.00 | |
| 7 | o4 Mini High OpenAI | 25.3% | Not benchmarked | $4.40 | |
| 8 | Gemini 2.5 Pro Google | 23.3% | 37.0 | $10.00 | |
| 9 | GPT-4o OpenAI | 9.9% | 38.8 | $10.00 |
How this list is built
Generated from the dataset on every refresh. Nobody picks the order by hand.
Models by their GDPval result, as the board publishes it, best first. Work products from 44 occupations, judged against the work of professionals. The number is the share of comparisons the model won or tied.
- OpenRouter, fetched
- Epoch AI benchmarking hub, fetched