Rankings / Anthropic
Claude Sonnet 4.6
released 2026-02-17
BenchAtlas Index
as of 2026-09-11
max effort
rank #100
8 families · 4 categories · medium
max effort
rank #105
8 families · 4 categories · medium
32K budget
rank #140
4 families · 3 categories · medium
Benchmark evidence
105 results
agentic coding
| Release 2026-06-25 tasks_counted=3 · livebench_version=2026-06-25 | medium effort | 42.6 % | independent | LiveBench |
| τ²-Bench | high effort | 79.5 % | independent | Artificial Analysis |
| τ²-Bench | low effort | 79.0 % | independent | Artificial Analysis |
| τ²-Bench | max effort | 75.7 % | independent | Artificial Analysis |
| τ²-Bench subset=banking | max effort | 34.4 % | independent | Artificial Analysis |
coding
| Release 2026-06-25 tasks_counted=2 · livebench_version=2026-06-25 | medium effort | 79.3 % | independent | LiveBench |
| SciCode | max effort | 50.1 % | independent | Artificial Analysis |
| SciCode | high effort | 46.9 % | independent | Artificial Analysis |
| SciCode | low effort | 44.1 % | independent | Artificial Analysis |
data analysis
| Release 2026-06-25 tasks_counted=3 · livebench_version=2026-06-25 | medium effort | 78.0 % | independent | LiveBench |
external indices
| AA Coding Index | max effort | 63.0 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | max effort | 30.5 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | high effort | 24.7 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | low effort | 23.3 points | independent | Artificial Analysis |
| ECI | medium effort | 152.2 points | independent | Epoch AI Benchmarking Hub |
| ECI | 152.2 points | independent | Epoch AI Benchmarking Hub | |
| ECI | 16K budget | 152.2 points | independent | Epoch AI Benchmarking Hub |
| ECI | 32K budget | 152.2 points | independent | Epoch AI Benchmarking Hub |
| ECI | max effort | 152.2 points | independent | Epoch AI Benchmarking Hub |
| ECI | low effort | 152.2 points | independent | Epoch AI Benchmarking Hub |
| ECI | high effort | 152.2 points | independent | Epoch AI Benchmarking Hub |
| Vals Index | max effort | 50.6 points | independent | Vals AI |
factuality
| SimpleQA Verified | high effort | 35.5 % | independent | Epoch AI Benchmarking Hub 2026-08-10 |
| SimpleQA Verified | max effort | 32.8 % | independent | Epoch AI Benchmarking Hub 2026-08-27 |
| SimpleQA Verified | 32K budget | 29.0 % | independent | Epoch AI Benchmarking Hub 2026-02-21 |
human preference
| Coding (style control)older version arena=text · category=coding · style_control=true | 1528.3 1522.6–1534.1 | community | LMArena Leaderboard Dataset | |
| Industry Software And It Services (style control)older version arena=text · category=industry_software_and_it_services · style_control=true | 1514.9 1509.8–1520.0 | community | LMArena Leaderboard Dataset | |
| Expert (style control)older version arena=text · category=expert · style_control=true | 1508.2 1499.9–1516.4 | community | LMArena Leaderboard Dataset | |
| Hard Prompts English (style control)older version arena=text · category=hard_prompts_english · style_control=true | 1506.9 1501.3–1512.6 | community | LMArena Leaderboard Dataset | |
| Chinese (style control)older version arena=text · category=chinese · style_control=true | 1506.1 1495.7–1516.5 | community | LMArena Leaderboard Dataset | |
| Hard Prompts (style control)older version arena=text · category=hard_prompts · style_control=true | 1503.8 1499.4–1508.2 | 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 | 1500.7 1493.9–1507.4 | community | LMArena Leaderboard Dataset | |
| Longer Query (style control)older version arena=text · category=longer_query · style_control=true | 1493.4 1488.2–1498.5 | community | LMArena Leaderboard Dataset | |
| Industry Medicine And Healthcare (style control)older version arena=text · category=industry_medicine_and_healthcare · style_control=true | 1491.2 1482.0–1500.5 | community | LMArena Leaderboard Dataset | |
| Industry Legal And Government (style control)older version arena=text · category=industry_legal_and_government · style_control=true | 1485.7 1476.7–1494.7 | community | LMArena Leaderboard Dataset | |
| Industry Mathematical (style control)older version arena=text · category=industry_mathematical · style_control=true | 1483.9 1473.2–1494.6 | 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 | 1483.5 1477.2–1489.9 | community | LMArena Leaderboard Dataset | |
| English (style control)older version arena=text · category=english · style_control=true | 1483.1 1478.3–1487.9 | community | LMArena Leaderboard Dataset | |
| Exclude Ties (style control)older version arena=text · category=exclude_ties · style_control=true | 1482.3 1477.5–1487.0 | community | LMArena Leaderboard Dataset | |
| Multi Turn (style control)older version arena=text · category=multi_turn · style_control=true | 1480.7 1474.1–1487.3 | community | LMArena Leaderboard Dataset | |
| French (style control)older version arena=text · category=french · style_control=true | 1479.6 1464.8–1494.4 | community | LMArena Leaderboard Dataset | |
| Instruction Following (style control)older version arena=text · category=instruction_following · style_control=true | 1476.1 1470.8–1481.5 | community | LMArena Leaderboard Dataset | |
| Overall (style control) arena=text · category=overall · style_control=true | 1472.4 1468.8–1476.0 | community | LMArena Leaderboard Dataset | |
| Polish (style control)older version arena=text · category=polish · style_control=true | 1471.0 1454.8–1487.2 | community | LMArena Leaderboard Dataset | |
| Spanish (style control)older version arena=text · category=spanish · style_control=true | 1467.9 1453.6–1482.3 | community | LMArena Leaderboard Dataset | |
| Russian (style control)older version arena=text · category=russian · style_control=true | 1465.7 1458.0–1473.4 | community | LMArena Leaderboard Dataset | |
| Math (style control)older version arena=text · category=math · style_control=true | 1463.0 1452.6–1473.4 | community | LMArena Leaderboard Dataset | |
| Non English (style control)older version arena=text · category=non_english · style_control=true | 1457.4 1453.0–1461.9 | 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 | 1456.2 1450.3–1462.0 | community | LMArena Leaderboard Dataset | |
| German (style control)older version arena=text · category=german · style_control=true | 1450.4 1432.0–1468.7 | community | LMArena Leaderboard Dataset | |
| Creative Writing (style control)older version arena=text · category=creative_writing · style_control=true | 1449.6 1442.8–1456.4 | 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 | 1445.7 1439.5–1452.0 | community | LMArena Leaderboard Dataset | |
| Japanese (style control)older version arena=text · category=japanese · style_control=true | 1438.5 1415.3–1461.8 | community | LMArena Leaderboard Dataset | |
| Korean (style control)older version arena=text · category=korean · style_control=true | 1434.9 1415.8–1453.9 | community | LMArena Leaderboard Dataset |
knowledge science
| GPQA Diamond implementation=artificial-analysis | max effort | 87.5 % | independent | Artificial Analysis |
| GPQA Diamond | 32K budget | 87.4 % | independent | Epoch AI Benchmarking Hub 2026-02-20 |
| GPQA Diamond implementation=vals-ai | max effort | 85.6 % | independent | Vals AI |
| GPQA Diamond | high effort | 83.3 % | independent | Epoch AI Benchmarking Hub 2026-07-13 |
| GPQA Diamond | medium effort | 83.3 % | independent | Epoch AI Benchmarking Hub 2026-07-13 |
| GPQA Diamond implementation=artificial-analysis | high effort | 79.9 % | independent | Artificial Analysis |
| GPQA Diamond implementation=artificial-analysis | low effort | 79.7 % | independent | Artificial Analysis |
| GPQA Diamond | max effort | 78.8 % | independent | Epoch AI Benchmarking Hub 2026-08-06 |
| Humanity's Last Exam implementation=artificial-analysis | max effort | 33.6 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | high effort | 13.3 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | low effort | 11.2 % | independent | Artificial Analysis |
| MMLU-Pro implementation=vals-ai | max effort | 87.3 % | independent | Vals AI |
language
| Release 2026-06-25 tasks_counted=3 · livebench_version=2026-06-25 | medium effort | 76.1 % | independent | LiveBench |
long context instruction
| AA-LCR | max effort | 80.0 % | independent | Artificial Analysis |
| AA-LCR | low effort | 69.3 % | independent | Artificial Analysis |
| AA-LCR | high effort | 68.3 % | independent | Artificial Analysis |
| IFBench | max effort | 56.6 % | independent | Artificial Analysis |
| IFBench | low effort | 42.4 % | independent | Artificial Analysis |
| IFBench | high effort | 41.2 % | independent | Artificial Analysis |
| Release 2026-06-25 tasks_counted=4 · livebench_version=2026-06-25 | medium effort | 63.2 % | independent | LiveBench |
multimodal
| MMMU implementation=vals-ai | max effort | 83.6 % | independent | Vals AI |
professional
| CaseLaw | 64.0 % | independent | Vals AI | |
| CorpFin | max effort | 65.3 % | independent | Vals AI |
| LegalBench | max effort | 82.1 % | independent | Vals AI |
| MedQA | thinking | 92.1 % | independent | Vals AI |
| TaxEval | max effort | 77.1 % | independent | Vals AI |
reasoning math
| AIME implementation=vals-ai | 92.3 % | independent | Vals AI | |
| ARC-AGI-1older version split=public_eval · model_type=CoT | max effort | 95.8 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | high effort | 95.3 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | high effort | 86.5 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | max effort | 86.0 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | high effort | 65.7 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | max effort | 62.4 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | high effort | 60.4 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | max effort | 58.3 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | high effort | 3.0 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | max effort | 2.9 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | max effort | 2.7 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | high effort | 2.7 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | max effort | 1.5 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | high effort | 1.2 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | max effort | 1.1 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | high effort | 0.8 usd_per_task | independent | ARC Prize Leaderboard |
| FrontierMath | 16K budget | 32.4 % | independent | Epoch AI Benchmarking Hub 2026-02-20 |
| FrontierMath Tier 4older version | 16K budget | 8.3 % | independent | Epoch AI Benchmarking Hub 2026-02-21 |
| Release 2026-06-25 tasks_counted=4 · livebench_version=2026-06-25 | medium effort | 87.0 % | independent | LiveBench |
| Release 2026-06-25 tasks_counted=4 · livebench_version=2026-06-25 | medium effort | 84.8 % | independent | LiveBench |
| OTIS Mock AIME 2024–2025 | 32K budget | 85.8 % | independent | Epoch AI Benchmarking Hub 2026-02-20 |
| OTIS Mock AIME 2024–2025 | medium effort | 82.2 % | independent | Epoch AI Benchmarking Hub 2026-07-13 |
| OTIS Mock AIME 2024–2025 | high effort | 75.6 % | independent | Epoch AI Benchmarking Hub 2026-07-13 |
| OTIS Mock AIME 2024–2025 | max effort | 71.1 % | independent | Epoch AI Benchmarking Hub 2026-08-06 |
Agent + model results
systems, not bare-model scores
| agent + model Epoch Inspect harness + Claude Sonnet 4.6 | SWE-bench Verified | 75.2 % | independent | Epoch AI Benchmarking Hub |
| agent + model Artificial Analysis harness + Claude Sonnet 4.6 | Terminal-Bench 2.1 | 71.2 % | independent | Artificial Analysis |
| agent + model terminal-agent + Claude Sonnet 4.6 | Terminal-Bench 2.0 | 53.4 % | unverified | Terminal-Bench Leaderboard |
| agent + model Artificial Analysis harness + Claude Sonnet 4.6 | Terminal-Bench Hard | 53.0 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + Claude Sonnet 4.6 | Terminal-Bench Hard | 46.2 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + Claude Sonnet 4.6 | Terminal-Bench Hard | 42.4 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.