GPT-4.1 vs GPT-o3 in finance

Every finance benchmark where both have a score (OpenAI and OpenAI). Data refreshed 2 Nov 2026.

Shared benchmarks, side by side

Each benchmark is drawn on its own scale. Bars are not comparable across rows.

CAIA (Crypto AI Agent Benchmark) accuracy % (majority vote, 5 runs), higher is better
GPT-4.1:46.6
GPT-o3:53.4 (better)
LiveTradeBench cumulative return %, higher is better
GPT-4.1:6.3 (better)
GPT-o3:6.0

Benchmark

GPT-4.1

GPT-o3

4 benchmarks, 2 scored for both. Bold marks the better score where both have one.

CAIA (Crypto AI Agent Benchmark)accuracy % (majority vote, 5 runs), higher is better

46.6
Paper Tables 2-3 (Reported by the authors, 17 Jan 2026)

53.4
Paper Tables 2-3 (Reported by the authors, 17 Jan 2026)

LiveTradeBenchcumulative return %, higher is better

6.3
Paper Table 2 (Reported by the authors, 5 Nov 2025)

6.0
Paper Table 2 (Reported by the authors, 5 Nov 2025)

FAITHoverall accuracy %, higher is better

89.2
Paper Table 3 (Reported by the authors, 24 Oct 2025)

Not scored

FinTrustsafety harmfulness score (1-10, GPT-4.1 mini judge), lower is better

7.4
Paper Table 7 (App. A.2) (Reported by the authors, 17 Oct 2025)

Not scored

Scores from different benchmarks use different scales and are not added up. Settings can differ between sources; see each leaderboard for details.