GPT-5.4
Agent-ready GPT for coding and computer-use workflows at a lower cost
Shogo Agent Index
Supported#33 of 9251.6range 47.2–56.1 · 64% of weight measured
Agentic coding
57
74% measured
Long-horizon autonomy
61
100% measured
Knowledge-work agents
40
52% measured
Computer use
42
estimated
Security & context learning
54
50% measured
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | 76.9% | 8% | source | |
| Terminal-Bench 2.0 | 81.8% | 8% | ForgeCode | |
| DeepSWE | 51.8% | 6% | mini-swe-agent · $5.65/task | |
| FrontierSWE | est. | 5% | predicted 23 | |
| CursorBench | est. | 4% | predicted 41 | |
| GSO-Bench | 31.4% | 4% | OpenHands | |
| METR Time Horizons | 74.3% | 10% | source | |
| Vending-Bench 2 | $6,144 | 6% | source | |
| PostTrainBench | 17.8% | 4% | Codex CLI | |
| APEX-Agents | 52.4% | 8% | source | |
| GDPval | est. | 7% | predicted 53 | |
| Remote Labor Index | est. | 5% | predicted 6 | |
| DeepResearch Bench | 35.1% | 5% | source | |
| OSWorld 2.0 | est. | 6% | predicted 14 | |
| OSWorld | est. | 4% | predicted 83 | |
| Cybench | est. | 5% | predicted 81 | |
| CL-bench | 27.9% | 5% | source |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| AlgoTune | Coding | 1.85× | source |
| ARC-AGI | Reasoning | 93.7% | source |
| ARC-AGI-2 | Reasoning | 74.0% | source |
| Chess Puzzles | Reasoning | 48.0% | source |
| CritPt | Reasoning | 23.4% | source |
| DTBench | Knowledge | 94.4% | source |
| FrontierMath Tiers 1-3 | Math | 78.6% | source |
| FrontierMath Tier 4 | Math | 49.0% | source |
| GPQA Diamond | Reasoning | 93.3% | source |
| Humanity's Last Exam | Reasoning | 36.2% | source |
| LMCA | Knowledge | 52.0% | source |
| MirrorCode | Coding | 23.3% | source |
| OTIS Mock AIME | Math | 97.8% | source |
| ProofBench | Math | 56.0% | source |
| SciCode | Coding | 56.6% | source |
| SimpleQA Verified | Knowledge | 45.1% | source |
| WebDev Arena | Coding | 1463 | source |