gpt-5
Benchmarks scored7
- CAIA (Crypto AI Agent Benchmark)Live market data · accuracy % (majority vote, 5 runs) · 17 Jan 2026
- 67.4#1 of 10
- K-FinHalluVerifiability · binary detection F1 · 28 May 2026
- 0.823#3 of 10
- LiveTradeBenchTrading and portfolios · cumulative return % · 5 Nov 2025
- 5.3#3 of 10
- Reasoning Models Ace the CFA Exams (CFA mock-exam evaluation)Finance knowledge · Level I accuracy % · 9 Dec 2025
- 96.1#2 of 6
- AI-TraderTrading and portfolios · cumulative return % · 1 Dec 2025
- 1.6#4 of 6
- PRBench FinanceAdvice and safety · weighted rubric score (0-100) · date not stated
- 51.3#7 of 10
- StockBenchTrading and portfolios · final return % · 2 Mar 2026
- 0.300#9 of 11