GPT-5.4 (xHigh): benchmark results
GPT-5.4 evaluated at the xhigh reasoning-effort setting. Provider: OpenAI. Released 2026-03-05. Access: API.
Aggregate Rating 1816 ± 7, rank #107 of 2168 rated models, from 246 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CLBench | 27.9 | Solving Rate (%) | 100 |
| CLBench - Domain Knowledge Reasoning | 28.8 | Solving Rate (%) | 100 |
| CLBench - Empirical Discovery & Simulation | 22.8 | Solving Rate (%) | 100 |
| CLBench - Procedural Task Execution | 30.1 | Solving Rate (%) | 100 |
| CLBench - Rule System Application | 26.7 | Solving Rate (%) | 100 |
| CiteVQA - Answer Correctness | 87.1 | Answer correctness against the reference answer, judged on a | 100 |
| Excel-Bench (Humanlaya) | 72.92 | Average Score (%) | 100 |
| FrontierFinance (Financial Modeling) | 87.5 | Average rubric score (%) over all 25 tasks (five each of DCF | 100 |
| FrontierFinance (Financial Modeling) - DCF | 97.4 | Average rubric score (%) over the five discounted cash flow | 100 |
| FrontierFinance (Financial Modeling) - Expert-Graded | 70.9 | Average rubric score (%) over ten tasks (two per model type) | 100 |
| FrontierFinance (Financial Modeling) - LBO | 85.7 | Average rubric score (%) over the five leveraged buyout mode | 100 |
| FrontierFinance (Financial Modeling) - Lender | 85.5 | Average rubric score (%) over the five lender model tasks, j | 100 |
Interactive version: theaggregate.ai/model?slug=gpt-5-4-xhigh · How It Works · Data refreshed daily, snapshot 2026-10-20.