GPT-5 vs GPT-5.4
Head to head on 19 shared benchmarks, with price, context window, and release dates.
Benchmark scores
82evals · sort by any column, ⤢ to expand fullscreen
| Model | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GPT-5.4 OpenAI | 65.3% | 1836elo | 78.3 | 47.6% | 87.2% | 74.8% | 11.3% | 48.4% | 78.5% | 41.3% | 77.5% | 68.3% | 56.0% | 43.3% | 47.1% | 74.0% | 37.9% | 67.4% | 36.0% | - | - | - | - | - | - | 13.1% | - | - | - | - | 1.47 | 81.0% | 34.1% | 9.0% | - | - | - | 1272elo | - | - | - | - | - | 92063 | - | 82.0% | 77.5% | 79.3% | 70.2% | 82.6% | 94.1% | 88.1% | - | - | - | - | - | - | 82.8% | - | - | - | - | - | - | - | 86 | - | - | 52.2% | - | -0.61 | 40.8 | - | - | - | 80 | - | - | - | - | - | 60.3% |
| GPT-5 OpenAI | 61.1% | 1623elo | 72.8 | 32.4% | 83.3% | 68.6% | 6.6% | 45.0% | 83.4% | 49.6% | 83.7% | 65.5% | 18.0% | 43.7% | 37.8% | 73.4% | 12.9% | 20.1% | 0.0% | 46.7% | 91.1% | 88.0% | 94.6% | 48.3% | 0.0% | - | 0.15 | 0.0% | 68.1% | 4.58 | - | - | - | - | 57.7% | 80.0% | 80.0% | - | 91.7% | 0.0% | 88.9% | 100.0% | 48.9% | - | 58.2% | - | - | - | - | - | - | - | 54.3% | 37.5% | 0.52 | 1.62 | 100.0% | 86.1% | - | 0.0% | 92.7% | 92.7% | 93.3% | 10.0% | 82.0% | 0.0% | - | 60.0% | 0.26 | - | 0.0% | - | - | 74.9% | 74.4% | 2.34 | - | 87.5% | 0.86 | 54.8% | 90.0% | 90.0% | 56.7% |
2 / 2 models
| Attribute | GPT-5 OpenAI | GPT-5.4 OpenAI |
|---|---|---|
| Description | OpenAI's August 2025 unified frontier model that auto-routes between a fast model and a deeper "thinking" variant. | GPT-5.4 is an AI model from OpenAI. |
| Family | GPT 5 | GPT |
| Params | Undisclosed (closed) | - |
| Open weights | closed | closed |
| License | Proprietary | proprietary |
| Released | 7 Aug 2025 | 5 Mar 2026 |
| Context | 400,000 | 1,050,000 |
| Input $/1M | - | $2.50 |
| Output $/1M | - | $15.00 |
| Cache read $/1M | $0.125 | $0.250 |
| Cache write $/1M | - | - |
| Speed | 0 tok/s | 0 tok/s |
| Latency | 0.00 s | 0.00 s |
| Modalities | text, image, file | text, image, file |
| Providers | openai | - |
| Publisher | OpenAI | OpenAI |
| Reported scores | GPQA Diamond74.8% (AA) SciCode47.1% (AA) τ²-bench (Tau²-bench)36.0% (AA) Terminal-Bench (Hard)37.9% (AA) Vibe Code Bench v1.167.4% |
FAQ
- Is GPT-5 or GPT-5.4 better?
- Across the 19 benchmarks both models report on Sophon, GPT-5.4 leads on 15 of them. Which one is "better" depends on the benchmark - the table above shows every shared score.
- How do GPT-5 and GPT-5.4 compare on GPQA Diamond?
- GPT-5 scores 68.6% and GPT-5.4 scores 74.8% on GPQA Diamond.
- When were GPT-5 and GPT-5.4 released?
- GPT-5 was released 7 Aug 2025 by OpenAI. GPT-5.4 was released 5 Mar 2026 by OpenAI.
Comparing something else? Build your own side-by-side.