Rankings / OpenAI
GPT 5 (2025-08-07)
proprietaryreleased 2025-08-07
BenchAtlas Index
as of 2026-09-11
high effort
rank #92
10 families · 5 categories · high
high effort
rank #103
15 families · 5 categories · high
medium effort
rank #124
15 families · 5 categories · high
Benchmark evidence
187 results
agentic coding
| τ²-Bench | medium effort | 86.5 % | independent | Artificial Analysis |
| τ²-Bench | high effort | 84.8 % | independent | Artificial Analysis |
| τ²-Bench | low effort | 84.2 % | independent | Artificial Analysis |
| τ²-Bench | minimal effort | 67.0 % | independent | Artificial Analysis |
| τ²-Bench subset=banking | high effort | 22.1 % | independent | Artificial Analysis |
| τ²-Bench | 0.0 % | independent | Artificial Analysis |
coding
| LiveCodeBench v6 implementation=artificial-analysis | high effort | 84.6 % | independent | Artificial Analysis |
| LiveCodeBench v6 implementation=artificial-analysis | low effort | 76.3 % | independent | Artificial Analysis |
| LiveCodeBench v6 implementation=artificial-analysis | medium effort | 70.3 % | independent | Artificial Analysis |
| LiveCodeBench v6 implementation=artificial-analysis | minimal effort | 55.8 % | independent | Artificial Analysis |
| LiveCodeBench v6 implementation=artificial-analysis | 54.3 % | independent | Artificial Analysis | |
| SciCode | high effort | 42.9 % | independent | Artificial Analysis |
| SciCode | medium effort | 41.1 % | independent | Artificial Analysis |
| SciCode | low effort | 39.1 % | independent | Artificial Analysis |
| SciCode | minimal effort | 38.8 % | independent | Artificial Analysis |
| SciCode | 37.8 % | independent | Artificial Analysis |
external indices
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. | 79.3 % 74.0–84.6 | independent | Scale Labs | |
| SimpleQA Verified | high effort | 50.6 % | independent | Epoch AI Benchmarking Hub 2025-12-09 |
| SimpleQA Verified | high effort | 50.1 % | independent | Epoch AI Benchmarking Hub 2026-08-27 |
human preference
| Coding (style control)older version arena=text · category=coding · style_control=true | 1470.0 1461.9–1478.0 | community | LMArena Leaderboard Dataset | |
| Coding (style control)older version arena=text · category=coding · style_control=true | high effort | 1467.6 1459.5–1475.6 | community | LMArena Leaderboard Dataset |
| Expert (style control)older version arena=text · category=expert · style_control=true | 1460.0 1444.6–1475.4 | community | LMArena Leaderboard Dataset | |
| Expert (style control)older version arena=text · category=expert · style_control=true | high effort | 1459.4 1443.6–1475.2 | community | LMArena Leaderboard Dataset |
| Industry Medicine And Healthcare (style control)older version arena=text · category=industry_medicine_and_healthcare · style_control=true | 1458.1 1443.2–1473.0 | community | LMArena Leaderboard Dataset | |
| Industry Legal And Government (style control)older version arena=text · category=industry_legal_and_government · style_control=true | 1457.8 1443.4–1472.2 | community | LMArena Leaderboard Dataset | |
| Industry Medicine And Healthcare (style control)older version arena=text · category=industry_medicine_and_healthcare · style_control=true | high effort | 1455.8 1440.9–1470.7 | community | LMArena Leaderboard Dataset |
| Industry Legal And Government (style control)older version arena=text · category=industry_legal_and_government · style_control=true | high effort | 1454.5 1440.2–1468.8 | community | LMArena Leaderboard Dataset |
| Chinese (style control)older version arena=text · category=chinese · style_control=true | high effort | 1454.0 1438.2–1469.9 | community | LMArena Leaderboard Dataset |
| Hard Prompts English (style control)older version arena=text · category=hard_prompts_english · style_control=true | 1453.9 1446.0–1461.7 | community | LMArena Leaderboard Dataset | |
| Chinese (style control)older version arena=text · category=chinese · style_control=true | 1453.5 1440.2–1466.8 | community | LMArena Leaderboard Dataset | |
| Industry Software And It Services (style control)older version arena=text · category=industry_software_and_it_services · style_control=true | 1453.4 1446.9–1460.0 | community | LMArena Leaderboard Dataset | |
| Hard Prompts English (style control)older version arena=text · category=hard_prompts_english · style_control=true | high effort | 1452.8 1445.3–1460.2 | 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 | 1452.4 1445.9–1458.9 | community | LMArena Leaderboard Dataset |
| Polish (style control)older version arena=text · category=polish · style_control=true | 1448.1 1433.0–1463.2 | community | LMArena Leaderboard Dataset | |
| Hard Prompts (style control)older version arena=text · category=hard_prompts · style_control=true | 1447.7 1441.8–1453.5 | community | LMArena Leaderboard Dataset | |
| Hard Prompts (style control)older version arena=text · category=hard_prompts · style_control=true | high effort | 1446.2 1440.3–1452.1 | community | LMArena Leaderboard Dataset |
| Polish (style control)older version arena=text · category=polish · style_control=true | high effort | 1445.8 1430.5–1461.1 | 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 | 1444.9 1436.1–1453.8 | community | LMArena Leaderboard Dataset | |
| German (style control)older version arena=text · category=german · style_control=true | 1444.7 1422.2–1467.3 | 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 | 1443.2 1434.3–1452.1 | community | LMArena Leaderboard Dataset |
| German (style control)older version arena=text · category=german · style_control=true | high effort | 1442.5 1419.2–1465.9 | community | LMArena Leaderboard Dataset |
| Industry Mathematical (style control)older version arena=text · category=industry_mathematical · style_control=true | high effort | 1442.3 1427.9–1456.6 | community | LMArena Leaderboard Dataset |
| Industry Mathematical (style control)older version arena=text · category=industry_mathematical · style_control=true | 1442.0 1427.6–1456.4 | community | LMArena Leaderboard Dataset | |
| English (style control)older version arena=text · category=english · style_control=true | 1439.3 1433.3–1445.3 | community | LMArena Leaderboard Dataset | |
| English (style control)older version arena=text · category=english · style_control=true | high effort | 1438.3 1432.6–1444.1 | community | LMArena Leaderboard Dataset |
| Overall (style control) arena=text · category=overall · style_control=true | 1434.4 1429.9–1438.9 | community | LMArena Leaderboard Dataset | |
| Math (style control)older version arena=text · category=math · style_control=true | high effort | 1434.2 1420.5–1447.8 | community | LMArena Leaderboard Dataset |
| Overall (style control) arena=text · category=overall · style_control=true | high effort | 1433.9 1429.4–1438.4 | community | LMArena Leaderboard Dataset |
| Math (style control)older version arena=text · category=math · style_control=true | 1433.5 1419.8–1447.2 | community | LMArena Leaderboard Dataset | |
| Japanese (style control)older version arena=text · category=japanese · style_control=true | 1430.5 1408.3–1452.7 | community | LMArena Leaderboard Dataset | |
| Japanese (style control)older version arena=text · category=japanese · style_control=true | high effort | 1429.7 1406.6–1452.9 | community | LMArena Leaderboard Dataset |
| Exclude Ties (style control)older version arena=text · category=exclude_ties · style_control=true | 1427.3 1421.2–1433.4 | community | LMArena Leaderboard Dataset | |
| Russian (style control)older version arena=text · category=russian · style_control=true | high effort | 1427.0 1412.2–1441.8 | community | LMArena Leaderboard Dataset |
| Russian (style control)older version arena=text · category=russian · style_control=true | 1426.9 1414.0–1439.8 | community | LMArena Leaderboard Dataset | |
| Exclude Ties (style control)older version arena=text · category=exclude_ties · style_control=true | high effort | 1426.8 1420.7–1432.9 | community | LMArena Leaderboard Dataset |
| Non English (style control)older version arena=text · category=non_english · style_control=true | high effort | 1426.0 1420.3–1431.8 | community | LMArena Leaderboard Dataset |
| Non English (style control)older version arena=text · category=non_english · style_control=true | 1423.9 1418.3–1429.5 | community | LMArena Leaderboard Dataset | |
| Multi Turn (style control)older version arena=text · category=multi_turn · style_control=true | 1420.8 1411.8–1429.8 | community | LMArena Leaderboard Dataset | |
| Multi Turn (style control)older version arena=text · category=multi_turn · style_control=true | high effort | 1420.5 1411.6–1429.4 | community | LMArena Leaderboard Dataset |
| Spanish (style control)older version arena=text · category=spanish · style_control=true | 1414.7 1393.9–1435.5 | community | LMArena Leaderboard Dataset | |
| Longer Query (style control)older version arena=text · category=longer_query · style_control=true | 1413.4 1405.7–1421.1 | 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 | 1412.9 1404.3–1421.5 | 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 | 1412.1 1403.5–1420.7 | community | LMArena Leaderboard Dataset |
| Longer Query (style control)older version arena=text · category=longer_query · style_control=true | high effort | 1411.5 1403.7–1419.3 | community | LMArena Leaderboard Dataset |
| Instruction Following (style control)older version arena=text · category=instruction_following · style_control=true | 1410.6 1403.4–1417.8 | community | LMArena Leaderboard Dataset | |
| Instruction Following (style control)older version arena=text · category=instruction_following · style_control=true | high effort | 1409.9 1402.7–1417.1 | community | LMArena Leaderboard Dataset |
| Spanish (style control)older version arena=text · category=spanish · style_control=true | high effort | 1403.7 1381.3–1426.1 | community | LMArena Leaderboard Dataset |
| Korean (style control)older version arena=text · category=korean · style_control=true | high effort | 1399.5 1376.5–1422.6 | 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 | 1398.0 1389.6–1406.3 | 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 | 1397.8 1389.4–1406.1 | 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 | 1397.1 1389.4–1404.8 | 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 | 1396.8 1389.1–1404.5 | community | LMArena Leaderboard Dataset | |
| Korean (style control)older version arena=text · category=korean · style_control=true | 1390.9 1368.5–1413.3 | community | LMArena Leaderboard Dataset | |
| Creative Writing (style control)older version arena=text · category=creative_writing · style_control=true | high effort | 1374.6 1364.9–1384.3 | community | LMArena Leaderboard Dataset |
| Creative Writing (style control)older version arena=text · category=creative_writing · style_control=true | 1373.4 1363.6–1383.1 | community | LMArena Leaderboard Dataset |
knowledge science
| GPQA Diamond | high effort | 86.2 % | independent | Epoch AI Benchmarking Hub 2025-10-29 |
| GPQA Diamond implementation=vals-ai | high effort | 85.6 % | independent | Vals AI |
| GPQA Diamond implementation=artificial-analysis | high effort | 85.4 % | independent | Artificial Analysis |
| GPQA Diamond | medium effort | 85.3 % | independent | Epoch AI Benchmarking Hub 2025-08-07 |
| GPQA Diamond implementation=artificial-analysis | medium effort | 84.2 % | independent | Artificial Analysis |
| GPQA Diamond implementation=artificial-analysis | low effort | 80.8 % | independent | Artificial Analysis |
| GPQA Diamond | minimal effort | 71.7 % | independent | Epoch AI Benchmarking Hub 2026-07-20 |
| GPQA Diamond implementation=artificial-analysis | 68.6 % | independent | Artificial Analysis | |
| GPQA Diamond implementation=artificial-analysis | minimal effort | 67.3 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | high effort | 28.5 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | medium effort | 25.4 % | 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. Sampled at reasoning_effort: 'high'. · implementation=scale | 25.3 % 23.6–27.0 | independent | Scale Labs | |
| Humanity's Last Exam implementation=artificial-analysis | low effort | 19.6 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | 6.6 % | independent | Artificial Analysis | |
| Humanity's Last Exam implementation=artificial-analysis | minimal effort | 6.0 % | independent | Artificial Analysis |
| MMLU-Pro implementation=artificial-analysis | high effort | 87.1 % | independent | Artificial Analysis |
| MMLU-Pro implementation=artificial-analysis | medium effort | 86.7 % | independent | Artificial Analysis |
| MMLU-Pro implementation=vals-ai | high effort | 86.5 % | independent | Vals AI |
| MMLU-Pro implementation=artificial-analysis | low effort | 86.0 % | independent | Artificial Analysis |
| MMLU-Pro implementation=artificial-analysis | 82.0 % | independent | Artificial Analysis | |
| MMLU-Pro implementation=artificial-analysis | minimal effort | 80.6 % | independent | Artificial Analysis |
long context instruction
| AA-LCR | high effort | 78.2 % | independent | Artificial Analysis |
| AA-LCR | medium effort | 76.0 % | independent | Artificial Analysis |
| AA-LCR | 65.0 % | independent | Artificial Analysis | |
| AA-LCR | low effort | 63.7 % | independent | Artificial Analysis |
| AA-LCR | minimal effort | 27.0 % | independent | Artificial Analysis |
| IFBench | high effort | 73.1 % | independent | Artificial Analysis |
| IFBench | medium effort | 70.6 % | independent | Artificial Analysis |
| IFBench | low effort | 66.6 % | independent | Artificial Analysis |
| IFBench | minimal effort | 45.6 % | independent | Artificial Analysis |
| IFBench | 45.0 % | independent | Artificial Analysis | |
| MultiChallenge | thinking | 63.2 % 61.6–64.8 | independent | Scale Labs |
multimodal
| MMMU implementation=vals-ai | high effort | 81.5 % | independent | Vals AI |
| VISTA | 49.7 % 48.7–50.7 | independent | Scale Labs |
professional
| CaseLaw | high effort | 66.5 % | independent | Vals AI |
| CorpFin | high effort | 61.1 % | independent | Vals AI |
| LegalBench | high effort | 86.0 % | independent | Vals AI |
| MedQA | high effort | 96.3 % | independent | Vals AI |
| TaxEval | high effort | 73.4 % | independent | Vals AI |
reasoning math
| AIME implementation=artificial-analysis | high effort | 95.7 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | high effort | 94.3 % | independent | Artificial Analysis |
| AIME implementation=vals-ai | high effort | 93.4 % | independent | Vals AI |
| AIME year=2025 · implementation=artificial-analysis | medium effort | 91.7 % | independent | Artificial Analysis |
| AIME implementation=artificial-analysis | medium effort | 91.7 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | low effort | 83.0 % | independent | Artificial Analysis |
| AIME implementation=artificial-analysis | low effort | 83.0 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | 48.3 % | independent | Artificial Analysis | |
| AIME implementation=artificial-analysis | minimal effort | 36.7 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | minimal effort | 31.7 % | independent | Artificial Analysis |
| ARC-AGI-1older version split=public_eval · model_type=CoT | high effort | 65.9 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | high effort | 65.7 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | medium effort | 63.4 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | medium effort | 56.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | low effort | 48.4 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | low effort | 44.0 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=Base LLM | minimal effort | 11.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | high effort | 9.9 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | high effort | 9.6 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | medium effort | 7.6 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | medium effort | 7.5 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=Base LLM | minimal effort | 6.0 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | low effort | 2.5 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | low effort | 1.9 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | high effort | 0.8 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | high effort | 0.7 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | high effort | 0.5 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | medium effort | 0.5 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | medium effort | 0.4 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | high effort | 0.4 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-1older version split=public_eval · model_type=CoT | medium effort | 0.2 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | low effort | 0.2 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | low effort | 0.2 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | low effort | 0.2 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 |
| ARC-AGI-2 split=semi_private · model_type=Base LLM | minimal effort | 0.1 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=Base LLM | minimal effort | 0.1 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=Base LLM | minimal effort | 0.0 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=Base LLM | minimal effort | 0.0 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=Base LLM | minimal effort | 0.0 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=Base LLM | minimal effort | 0.0 % | 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. | 10.5 % 8.7–12.2 | independent | Scale Labs | |
| FrontierMath | high effort | 32.4 % | independent | Epoch AI Benchmarking Hub 2025-11-13 |
| FrontierMath | medium effort | 27.2 % | independent | Epoch AI Benchmarking Hub 2025-11-13 |
| FrontierMath Tier 4older version | high effort | 12.5 % | independent | Epoch AI Benchmarking Hub 2025-10-30 |
| FrontierMath Tier 4older version | medium effort | 6.3 % | independent | Epoch AI Benchmarking Hub 2025-08-07 |
| MATH-500 implementation=artificial-analysis | high effort | 99.4 % | independent | Artificial Analysis |
| MATH-500 implementation=artificial-analysis | medium effort | 99.1 % | independent | Artificial Analysis |
| MATH-500 implementation=artificial-analysis | low effort | 98.7 % | independent | Artificial Analysis |
| MATH-500 implementation=vals-ai | high effort | 96.0 % | independent | Vals AI |
| MATH-500 implementation=artificial-analysis | minimal effort | 86.1 % | independent | Artificial Analysis |
| MATH Level 5 | high effort | 98.1 % | independent | Epoch AI Benchmarking Hub 2025-10-29 |
| MATH Level 5 | medium effort | 97.9 % | independent | Epoch AI Benchmarking Hub 2025-08-20 |
| OTIS Mock AIME 2024–2025 | high effort | 91.4 % | independent | Epoch AI Benchmarking Hub 2025-10-29 |
| OTIS Mock AIME 2024–2025 | medium effort | 87.2 % | independent | Epoch AI Benchmarking Hub 2025-08-07 |
| OTIS Mock AIME 2024–2025 | minimal effort | 46.7 % | independent | Epoch AI Benchmarking Hub 2026-07-20 |
Agent + model results
systems, not bare-model scores
| agent + model Prometheus-v1.2.1 + GPT 5 (2025-08-07) | SWE-bench Verified | 74.4 % | unverified | SWE-bench Leaderboard |
| agent + model Epoch Inspect harness + GPT 5 (2025-08-07) | SWE-bench Verified | 73.5 % | independent | Epoch AI Benchmarking Hub |
| agent + model Epoch Inspect harness + GPT 5 (2025-08-07) | SWE-bench Verified | 71.5 % | independent | Epoch AI Benchmarking Hub |
| agent + model Prometheus-v1.2 + GPT 5 (2025-08-07) | SWE-bench Verified | 71.2 % | unverified | SWE-bench Leaderboard |
| agent + model mini-SWE-agent + GPT 5 (2025-08-07) | SWE-bench bash-only | 65.0 % | community | SWE-bench Leaderboard |
| agent + model mini-SWE-agent + GPT 5 (2025-08-07) | SWE-bench Verified | 65.0 % | community | SWE-bench Leaderboard |
| agent + model mini-SWE-agent + GPT 5 (2025-08-07) | SWE-bench bash-only | 0.3 usd_per_task | community | SWE-bench Leaderboard |
| agent + model mini-SWE-agent + GPT 5 (2025-08-07) | SWE-bench Verified | 0.3 usd_per_task | community | SWE-bench Leaderboard |
| agent + model Droid + GPT 5 (2025-08-07) | Terminal-Bench 1.0 | 52.5 % | unverified | Terminal-Bench Leaderboard |
| agent + model Codex CLI + GPT 5 (2025-08-07) | Terminal-Bench 2.0 | 49.6 % | community | Terminal-Bench Leaderboard |
| agent + model apex2 + GPT 5 (2025-08-07) | Terminal-Bench 1.0 | 49.3 % | unverified | Terminal-Bench Leaderboard |
| agent + model OpenHands + GPT 5 (2025-08-07) | Terminal-Bench 2.0 | 43.8 % | community | Terminal-Bench Leaderboard |
| agent + model terminus-2 + GPT 5 (2025-08-07) | Terminal-Bench 1.0 | 41.3 % | community | Terminal-Bench Leaderboard |
| agent + model Artificial Analysis harness + GPT 5 (2025-08-07) | Terminal-Bench Hard | 37.9 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + GPT 5 (2025-08-07) | Terminal-Bench 2.1 | 35.2 % | independent | Artificial Analysis |
| agent + model terminus-2 + GPT 5 (2025-08-07) | Terminal-Bench 2.0 | 35.2 % | community | Terminal-Bench Leaderboard |
| agent + model mini-SWE-agent + GPT 5 (2025-08-07) | Terminal-Bench 2.0 | 33.9 % | community | Terminal-Bench Leaderboard |
| agent + model Artificial Analysis harness + GPT 5 (2025-08-07) | Terminal-Bench Hard | 32.6 % | independent | Artificial Analysis |
| agent + model terminus-1 + GPT 5 (2025-08-07) | Terminal-Bench 1.0 | 30.0 % | community | Terminal-Bench Leaderboard |
| agent + model Artificial Analysis harness + GPT 5 (2025-08-07) | Terminal-Bench Hard | 26.5 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + GPT 5 (2025-08-07) | Terminal-Bench Hard | 18.2 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + GPT 5 (2025-08-07) | Terminal-Bench Hard | 12.9 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.