Rankings / OpenAI
GPT 5.4 (2026-03-05)
proprietaryreleased 2026-03-05
BenchAtlas Index
as of 2026-07-14
xhigh effort
rank #22
17 families · 5 categories · high
xhigh effort
rank #38
10 families · 4 categories · medium
low effort
rank #58
9 families · 5 categories · high
Benchmark evidence
145 results
agentic coding
| Release 2026-06-25 tasks_counted=3 · livebench_version=2026-06-25 | xhigh effort | 53.8 % | independent | LiveBench |
| τ²-Bench | xhigh effort | 87.1 % | independent | Artificial Analysis |
| τ²-Bench | low effort | 74.6 % | independent | Artificial Analysis |
| τ²-Bench | no reasoning | 35.1 % | independent | Artificial Analysis |
| τ²-Bench subset=banking | xhigh effort | 30.3 % | independent | Artificial Analysis |
coding
| Release 2026-06-25 tasks_counted=2 · livebench_version=2026-06-25 | xhigh effort | 77.5 % | independent | LiveBench |
| SciCode | xhigh effort | 56.6 % | independent | Artificial Analysis |
| SciCode | low effort | 50.3 % | independent | Artificial Analysis |
| SciCode | no reasoning | 47.1 % | independent | Artificial Analysis |
data analysis
| Release 2026-06-25 tasks_counted=3 · livebench_version=2026-06-25 | xhigh effort | 79.3 % | independent | LiveBench |
external indices
| AA Coding Index | xhigh effort | 71.1 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | xhigh effort | 51.4 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | low effort | 39.1 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | no reasoning | 27.7 points | independent | Artificial Analysis |
| ECI | high effort | 156.0 points | independent | Epoch AI Benchmarking Hub |
| ECI | no reasoning | 156.0 points | independent | Epoch AI Benchmarking Hub |
| ECI | 156.0 points | independent | Epoch AI Benchmarking Hub | |
| ECI | low effort | 156.0 points | independent | Epoch AI Benchmarking Hub |
| ECI | medium effort | 156.0 points | independent | Epoch AI Benchmarking Hub |
| ECI | xhigh effort | 156.0 points | independent | Epoch AI Benchmarking Hub |
factuality
| MASK contamination=Potential contamination warning: This model was evaluated after the public release of MASK, allowing model builder access to the prompts and solutions. | max effort | 89.7 % 88.9–90.4 | independent | Scale Labs |
| SimpleQA Verified | xhigh effort | 44.8 % | independent | Epoch AI Benchmarking Hub 2026-03-06 |
human preference
| Coding (style control)older version arena=text · category=coding · style_control=true | high effort | 1519.6 1513.4–1525.9 | community | LMArena Leaderboard Dataset |
| Expert (style control)older version arena=text · category=expert · style_control=true | high effort | 1518.9 1509.5–1528.3 | community | LMArena Leaderboard Dataset |
| Coding (style control)older version arena=text · category=coding · style_control=true | 1512.7 1506.5–1518.9 | community | LMArena Leaderboard Dataset | |
| Chinese (style control)older version arena=text · category=chinese · style_control=true | high effort | 1508.2 1495.8–1520.6 | community | LMArena Leaderboard Dataset |
| Industry Software And It Services (style control)older version arena=text · category=industry_software_and_it_services · style_control=true | high effort | 1504.7 1499.1–1510.2 | community | LMArena Leaderboard Dataset |
| Industry Mathematical (style control)older version arena=text · category=industry_mathematical · style_control=true | high effort | 1500.8 1489.0–1512.7 | community | LMArena Leaderboard Dataset |
| French (style control)older version arena=text · category=french · style_control=true | high effort | 1500.4 1484.3–1516.5 | community | LMArena Leaderboard Dataset |
| Industry Software And It Services (style control)older version arena=text · category=industry_software_and_it_services · style_control=true | 1499.7 1494.2–1505.2 | community | LMArena Leaderboard Dataset | |
| Hard Prompts (style control)older version arena=text · category=hard_prompts · style_control=true | high effort | 1497.8 1493.0–1502.6 | community | LMArena Leaderboard Dataset |
| Chinese (style control)older version arena=text · category=chinese · style_control=true | 1495.2 1483.2–1507.3 | community | LMArena Leaderboard Dataset | |
| Hard Prompts English (style control)older version arena=text · category=hard_prompts_english · style_control=true | high effort | 1494.9 1488.9–1500.8 | community | LMArena Leaderboard Dataset |
| Math (style control)older version arena=text · category=math · style_control=true | high effort | 1494.8 1483.3–1506.3 | community | LMArena Leaderboard Dataset |
| Expert (style control)older version arena=text · category=expert · style_control=true | 1494.0 1484.8–1503.3 | community | LMArena Leaderboard Dataset | |
| Russian (style control)older version arena=text · category=russian · style_control=true | high effort | 1492.1 1483.3–1500.8 | community | LMArena Leaderboard Dataset |
| Multi Turn (style control)older version arena=text · category=multi_turn · style_control=true | high effort | 1492.0 1484.9–1499.1 | community | LMArena Leaderboard Dataset |
| Industry Life And Physical And Social Science (style control)older version arena=text · category=industry_life_and_physical_and_social_science · style_control=true | high effort | 1488.6 1481.3–1496.0 | community | LMArena Leaderboard Dataset |
| Hard Prompts (style control)older version arena=text · category=hard_prompts · style_control=true | 1487.8 1483.0–1492.6 | community | LMArena Leaderboard Dataset | |
| Hard Prompts English (style control)older version arena=text · category=hard_prompts_english · style_control=true | 1487.7 1481.8–1493.6 | community | LMArena Leaderboard Dataset | |
| Industry Legal And Government (style control)older version arena=text · category=industry_legal_and_government · style_control=true | high effort | 1487.5 1477.5–1497.5 | community | LMArena Leaderboard Dataset |
| Exclude Ties (style control)older version arena=text · category=exclude_ties · style_control=true | high effort | 1485.8 1480.6–1491.1 | community | LMArena Leaderboard Dataset |
| French (style control)older version arena=text · category=french · style_control=true | 1483.7 1467.7–1499.7 | community | LMArena Leaderboard Dataset | |
| Industry Business And Management And Financial Operations (style control)older version arena=text · category=industry_business_and_management_and_financial_operations · style_control=true | high effort | 1483.6 1476.7–1490.6 | community | LMArena Leaderboard Dataset |
| Polish (style control)older version arena=text · category=polish · style_control=true | high effort | 1483.3 1465.3–1501.4 | community | LMArena Leaderboard Dataset |
| Longer Query (style control)older version arena=text · category=longer_query · style_control=true | high effort | 1482.9 1477.2–1488.6 | community | LMArena Leaderboard Dataset |
| Multi Turn (style control)older version arena=text · category=multi_turn · style_control=true | 1480.1 1473.1–1487.1 | community | LMArena Leaderboard Dataset | |
| English (style control)older version arena=text · category=english · style_control=true | high effort | 1478.8 1473.6–1484.0 | community | LMArena Leaderboard Dataset |
| Longer Query (style control)older version arena=text · category=longer_query · style_control=true | 1478.5 1472.8–1484.1 | community | LMArena Leaderboard Dataset | |
| Industry Life And Physical And Social Science (style control)older version arena=text · category=industry_life_and_physical_and_social_science · style_control=true | 1478.4 1471.2–1485.6 | community | LMArena Leaderboard Dataset | |
| Overall (style control) arena=text · category=overall · style_control=true | high effort | 1476.6 1472.5–1480.6 | community | LMArena Leaderboard Dataset |
| Industry Legal And Government (style control)older version arena=text · category=industry_legal_and_government · style_control=true | 1476.3 1466.5–1486.1 | community | LMArena Leaderboard Dataset | |
| Industry Medicine And Healthcare (style control)older version arena=text · category=industry_medicine_and_healthcare · style_control=true | high effort | 1476.2 1465.6–1486.7 | community | LMArena Leaderboard Dataset |
| Industry Business And Management And Financial Operations (style control)older version arena=text · category=industry_business_and_management_and_financial_operations · style_control=true | 1475.6 1468.8–1482.5 | community | LMArena Leaderboard Dataset | |
| Instruction Following (style control)older version arena=text · category=instruction_following · style_control=true | high effort | 1475.5 1469.7–1481.4 | community | LMArena Leaderboard Dataset |
| Industry Mathematical (style control)older version arena=text · category=industry_mathematical · style_control=true | 1474.0 1462.3–1485.6 | community | LMArena Leaderboard Dataset | |
| Japanese (style control)older version arena=text · category=japanese · style_control=true | high effort | 1473.3 1445.3–1501.3 | community | LMArena Leaderboard Dataset |
| Exclude Ties (style control)older version arena=text · category=exclude_ties · style_control=true | 1472.0 1466.8–1477.1 | community | LMArena Leaderboard Dataset | |
| Russian (style control)older version arena=text · category=russian · style_control=true | 1470.9 1462.4–1479.4 | community | LMArena Leaderboard Dataset | |
| Polish (style control)older version arena=text · category=polish · style_control=true | 1469.9 1451.8–1488.0 | community | LMArena Leaderboard Dataset | |
| English (style control)older version arena=text · category=english · style_control=true | 1469.9 1464.8–1475.0 | community | LMArena Leaderboard Dataset | |
| Non English (style control)older version arena=text · category=non_english · style_control=true | high effort | 1468.4 1463.4–1473.4 | community | LMArena Leaderboard Dataset |
| Industry Writing And Literature And Language (style control)older version arena=text · category=industry_writing_and_literature_and_language · style_control=true | high effort | 1465.8 1459.4–1472.2 | community | LMArena Leaderboard Dataset |
| Overall (style control) arena=text · category=overall · style_control=true | 1465.7 1461.7–1469.6 | community | LMArena Leaderboard Dataset | |
| Industry Medicine And Healthcare (style control)older version arena=text · category=industry_medicine_and_healthcare · style_control=true | 1465.5 1455.2–1475.7 | community | LMArena Leaderboard Dataset | |
| German (style control)older version arena=text · category=german · style_control=true | high effort | 1465.1 1444.4–1485.8 | community | LMArena Leaderboard Dataset |
| Spanish (style control)older version arena=text · category=spanish · style_control=true | 1462.5 1446.3–1478.8 | community | LMArena Leaderboard Dataset | |
| Instruction Following (style control)older version arena=text · category=instruction_following · style_control=true | 1461.1 1455.3–1466.9 | community | LMArena Leaderboard Dataset | |
| Math (style control)older version arena=text · category=math · style_control=true | 1459.9 1448.8–1471.1 | community | LMArena Leaderboard Dataset | |
| Spanish (style control)older version arena=text · category=spanish · style_control=true | high effort | 1457.6 1441.5–1473.6 | community | LMArena Leaderboard Dataset |
| Non English (style control)older version arena=text · category=non_english · style_control=true | 1457.5 1452.5–1462.4 | community | LMArena Leaderboard Dataset | |
| Industry Writing And Literature And Language (style control)older version arena=text · category=industry_writing_and_literature_and_language · style_control=true | 1457.0 1450.6–1463.4 | community | LMArena Leaderboard Dataset | |
| German (style control)older version arena=text · category=german · style_control=true | 1449.8 1427.8–1471.7 | community | LMArena Leaderboard Dataset | |
| Japanese (style control)older version arena=text · category=japanese · style_control=true | 1449.3 1422.0–1476.6 | community | LMArena Leaderboard Dataset | |
| Creative Writing (style control)older version arena=text · category=creative_writing · style_control=true | high effort | 1443.6 1436.0–1451.1 | community | LMArena Leaderboard Dataset |
| Industry Entertainment And Sports And Media (style control)older version arena=text · category=industry_entertainment_and_sports_and_media · style_control=true | high effort | 1442.0 1435.1–1449.0 | community | LMArena Leaderboard Dataset |
| Korean (style control)older version arena=text · category=korean · style_control=true | high effort | 1432.4 1410.7–1454.1 | community | LMArena Leaderboard Dataset |
| Creative Writing (style control)older version arena=text · category=creative_writing · style_control=true | 1432.2 1424.8–1439.6 | community | LMArena Leaderboard Dataset | |
| Industry Entertainment And Sports And Media (style control)older version arena=text · category=industry_entertainment_and_sports_and_media · style_control=true | 1431.0 1424.3–1437.8 | community | LMArena Leaderboard Dataset | |
| Korean (style control)older version arena=text · category=korean · style_control=true | 1423.5 1402.3–1444.7 | community | LMArena Leaderboard Dataset |
knowledge science
| GPQA Diamond | xhigh effort | 93.3 % | independent | Epoch AI Benchmarking Hub 2026-03-06 |
| GPQA Diamond implementation=artificial-analysis | xhigh effort | 92.0 % | independent | Artificial Analysis |
| GPQA Diamond implementation=vals-ai | xhigh effort | 91.7 % | independent | Vals AI |
| GPQA Diamond implementation=artificial-analysis | low effort | 87.1 % | independent | Artificial Analysis |
| GPQA Diamond implementation=artificial-analysis | no reasoning | 74.8 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | xhigh effort | 41.6 % | independent | Artificial Analysis |
| Humanity's Last Exam contamination=Potential contamination warning: This model was evaluated after the public release of HLE, allowing model builder access to the prompts and solutions. · implementation=scale | max effort | 36.2 % 34.4–38.1 | independent | Scale Labs |
| Humanity's Last Exam implementation=artificial-analysis | low effort | 28.9 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | no reasoning | 10.6 % | independent | Artificial Analysis |
| MMLU-Pro implementation=vals-ai | xhigh effort | 87.5 % | independent | Vals AI |
language
| Release 2026-06-25 tasks_counted=3 · livebench_version=2026-06-25 | xhigh effort | 82.6 % | independent | LiveBench |
long context instruction
| AA-LCR | xhigh effort | 74.0 % | independent | Artificial Analysis |
| AA-LCR | low effort | 67.3 % | independent | Artificial Analysis |
| AA-LCR | no reasoning | 47.3 % | independent | Artificial Analysis |
| IFBench | xhigh effort | 74.0 % | independent | Artificial Analysis |
| IFBench | low effort | 65.9 % | independent | Artificial Analysis |
| IFBench | no reasoning | 48.4 % | independent | Artificial Analysis |
| Release 2026-06-25 tasks_counted=4 · livebench_version=2026-06-25 | xhigh effort | 70.2 % | independent | LiveBench |
multimodal
| MMMU implementation=vals-ai | xhigh effort | 87.5 % | independent | Vals AI |
| VISTA | max effort | 50.9 % 49.1–52.6 | independent | Scale Labs |
professional
| CaseLaw | xhigh effort | 63.8 % | independent | Vals AI |
| CorpFin | xhigh effort | 65.3 % | independent | Vals AI |
| LegalBench | xhigh effort | 86.0 % | independent | Vals AI |
| MedQA | xhigh effort | 96.1 % | independent | Vals AI |
| TaxEval | xhigh effort | 74.0 % | independent | Vals AI |
reasoning math
| AIME implementation=vals-ai | xhigh effort | 96.7 % | independent | Vals AI |
| ARC-AGI-1older version split=public_eval · model_type=CoT | xhigh effort | 96.4 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | high effort | 95.6 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | xhigh effort | 93.7 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | high effort | 92.7 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | medium effort | 92.0 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | medium effort | 86.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | xhigh effort | 84.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | low effort | 80.0 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | high effort | 75.8 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | xhigh effort | 74.0 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | low effort | 68.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | high effort | 67.5 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | medium effort | 58.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | medium effort | 55.4 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | low effort | 29.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | low effort | 23.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | xhigh effort | 1.6 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | xhigh effort | 1.5 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | high effort | 1.1 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | high effort | 1.0 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | medium effort | 0.7 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | medium effort | 0.7 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | xhigh effort | 0.6 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | xhigh effort | 0.4 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | high effort | 0.4 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | low effort | 0.3 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | low effort | 0.3 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | high effort | 0.3 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | medium effort | 0.3 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-3older version split=semi_private · model_type=CoT | high effort | 0.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | medium effort | 0.2 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | low effort | 0.1 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | low effort | 0.1 usd_per_task | independent | ARC Prize Leaderboard |
| EnigmaEval contamination=Potential contamination warning: This model was evaluated after the public release of EnigmaEval, allowing model builder access to the prompts and solutions. | max effort | 16.0 % 13.9–18.1 | independent | Scale Labs |
| FrontierMath | xhigh effort | 47.6 % | independent | Epoch AI Benchmarking Hub 2026-03-06 |
| FrontierMath Tier 4older version | xhigh effort | 27.1 % | independent | Epoch AI Benchmarking Hub 2026-03-06 |
| Release 2026-06-25 tasks_counted=4 · livebench_version=2026-06-25 | xhigh effort | 94.2 % | independent | LiveBench |
| Release 2026-06-25 tasks_counted=4 · livebench_version=2026-06-25 | xhigh effort | 88.1 % | independent | LiveBench |
| OTIS Mock AIME 2024–2025 | xhigh effort | 95.3 % | independent | Epoch AI Benchmarking Hub 2026-03-06 |
Agent + model results
systems, not bare-model scores
| agent + model Epoch Inspect harness + GPT 5.4 (2026-03-05) | SWE-bench Verified | 76.9 % | independent | Epoch AI Benchmarking Hub |
| agent + model Artificial Analysis harness + GPT 5.4 (2026-03-05) | Terminal-Bench 2.1 | 78.3 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + GPT 5.4 (2026-03-05) | Terminal-Bench Hard | 57.6 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + GPT 5.4 (2026-03-05) | Terminal-Bench Hard | 43.2 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + GPT 5.4 (2026-03-05) | Terminal-Bench Hard | 37.9 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.