# GPT-5-Thinking vs Numa 1

Every finance benchmark where both have a score (OpenAI and Moneda). Data refreshed 2 Nov 2026. Scores from different benchmarks use different scales and are not added up.

| Benchmark | Metric | GPT-5-Thinking | GPT-5-Thinking source | Numa 1 | Numa 1 source | Ahead |
| --- | --- | --- | --- | --- | --- | --- |
| FinSearchComp | Global subset average accuracy % (higher is better) | 63.9 | Paper Table 5, 16 Sept 2025, https://arxiv.org/html/2509.13160v1 | 59.4 | Moneda run, 2026-11-01, 1 Nov 2026, https://benchmarks.finance/method/ | GPT-5-Thinking |
| FinSearchComp | Greater China subset average accuracy % (higher is better) | 46.4 | Paper Table 5, 16 Sept 2025, https://arxiv.org/html/2509.13160v1 | 57.1 | Moneda run, 2026-11-01, 1 Nov 2026, https://benchmarks.finance/method/ | Numa 1 |

Source: https://benchmarks.finance/compare/gpt-5-thinking-vs-numa-1/

Created and maintained by Moneda.