RLTasksby ulam.ai
Models / Anthropic

Anthropic / MODEL EVIDENCE

Claude Sonnet 4.6

61 benchmarks · 83 observations · 66 sources · released 2026-02-17

Scores at a glance

61 benchmarks · large dot: Claude Sonnet 4.6; small dots: closest models where reported

ITBench-AA ↗39.80%
35.2256.21

Claude Sonnet 4.6 (Max) · ITBench-AA · root-cause score ↗

GPQA Diamond ↗85.61 %
27.0295.45

Claude Sonnet 4.6 · GPQA Diamond · Accuracy ↗

4 more reported results

87.37% · Claude Sonnet 4.6 (32k thinking) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

83.33% · Claude Sonnet 4.6 (high) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

83.33% · Claude Sonnet 4.6 (medium) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

78.79% · Claude Sonnet 4.6 (max) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

OTIS Mock AIME 2024-2025 ↗85.79%
85.7985.79

Claude Sonnet 4.6 (32k thinking) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % ↗

3 more reported results

82.22% · Claude Sonnet 4.6 (medium) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗

75.56% · Claude Sonnet 4.6 (high) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗

71.11% · Claude Sonnet 4.6 (max) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗

GBAEval ↗48.76%
48.7648.76

Claude Sonnet 4.6 (unknown thinking) · GBAEval · Overall score · % ↗

CritPt ↗3.14%
032.29

Claude Sonnet 4.6 (max) · CritPt · Accuracy · % ↗

ExploitBench ↗23.6%
13.373.8

Claude Sonnet 4.6 (no thinking) · ExploitBench · Mean capability · % ↗

1 more reported result

22.7% · Claude Sonnet 4.6 (no thinking) · ExploitBench · Mean capability · % · Epoch AI · ExploitBench ↗

LMCA ↗46.51%
2.7868.26

Claude Sonnet 4.6 (max) · LMCA · Score · % ↗

DTBench ↗89.87%
41.6598.93

Claude Sonnet 4.6 (max) · DTBench · Accuracy · % ↗

GDP.pdf ↗18%
235

Claude Sonnet 4.6 (max) · GDP.pdf · GDP.pdf score · % ↗

SciCode ↗46.76%
1.2766.9

Claude Sonnet 4.6 (max) · SciCode · Score · % ↗

Closest models by score

Similar positions on shared benchmark leaderboards. Smaller gaps are closer.

17.2 percentile-point gap

20 of 33 shared benchmark comparisons

17.9 percentile-point gap

21 of 33 shared benchmark comparisons

How similarity is measured

Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.

Filter comparisons to this model →

Sources

Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.

  1. Blueprint-Bench 2 | Andon Labs ↗

    Andon Labs

  2. Arena leaderboard — WebDev evaluation ↗

    Arena

  3. APEX-Agents-AA Benchmark | Artificial Analysis ↗

    Artificial Analysis

  4. Global-MMLU-Lite Benchmark | Artificial Analysis ↗

    Artificial Analysis

  5. ITBench-AA Benchmark | Artificial Analysis ↗

    Artificial Analysis

  6. SciCode Benchmark | Artificial Analysis ↗

    Artificial Analysis

  7. Terminal-Bench Hard Benchmark | Artificial Analysis ↗

    Artificial Analysis

  8. OfficeQA Pro: An Enterprise Benchmark for End-to-End Grounded Reasoning — published result table ↗

    Opsahl-Ong, Krista, Singhvi, Arnav, Collins, Jasmine, Zhou, Ivan, Wang, Cindy, Baheti, Ashutosh et al. · 2026

    All 13 authors

    Opsahl-Ong, Krista, Singhvi, Arnav, Collins, Jasmine, Zhou, Ivan, Wang, Cindy, Baheti, Ashutosh, Oertell, Owen, Portes, Jacob, Havens, Sam, Elsen, Erich, Bendersky, Michael, Zaharia, Matei, Chen, Xing

  9. FrontierCode Leaderboard ↗

    Cognition

  10. DTBench — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  11. LMCA — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  12. DeepSWE ↗

    Datacurve

  13. APEX-Agents ↗

    Epoch AI

  14. ARC-AGI-1 ↗

    Epoch AI

  15. ARC-AGI-2 ↗

    Epoch AI

  16. Chess Puzzles ↗

    Epoch AI

  17. CritPt v1 ↗

    Epoch AI

  18. ExploitBench v0.1 ↗

    Epoch AI

  19. FrontierMath Tier 4 (v1, old) ↗

    Epoch AI

  20. FrontierMath Tiers 1-3 (v1, old) ↗

    Epoch AI

  21. GPQA Diamond ↗

    Epoch AI

  22. Mystery Game Puzzles ↗

    Epoch AI

  23. OTIS Mock AIME 2024-2025 ↗

    Epoch AI

  24. SimpleQA Verified ↗

    Epoch AI

  25. SWE-bench Verified ↗

    Epoch AI

  26. Terminal-Bench 2.0 ↗

    Epoch AI

  27. Vending Bench 2 — methodology and published results ↗

    Epoch AI · 2026

  28. WeirdML (v2) ↗

    Epoch AI

  29. AI benchmarks & capabilities ↗

    Epoch AI

  30. FutureSearch Evals ↗

    FutureSearch

  31. Claude Sonnet 4.6 (unknown thinking) — GBA Eval ↗

    GBA Eval

  32. DrugDiscoveryBench — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  33. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  34. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  35. SWE Atlas - Refactoring — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  36. SWE Atlas - Test Writing — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  37. OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments ↗

    Xie, Tianbao, Zhang, Danyang, Chen, Jixuan, Li, Xiaochuan, Zhao, Siheng, Cao, Ruisheng et al. · 2024

    All 17 authors

    Xie, Tianbao, Zhang, Danyang, Chen, Jixuan, Li, Xiaochuan, Zhao, Siheng, Cao, Ruisheng, Hua, Toh Jing, Cheng, Zhoujun, Shin, Dongchan, Lei, Fangyu, Liu, Yitao, Xu, Yiheng, Zhou, Shuyan, Savarese, Silvio, Xiong, Caiming, Zhong, Victor, Yu, Tao

  38. OSWorld 2.0: Benchmarking computer-use agents on long-horizon real-world tasks ↗

    OSWorld Team

  39. GDP.pdf Benchmark | Surge AI ↗

    Surge AI

  40. Terminal-Bench 2.0 leaderboard ↗

    Terminal-Bench Team

  41. AIME — evaluation methodology and results ↗

    Vals AI · 2026

  42. CaseLaw — evaluation methodology and results ↗

    Vals AI · 2026

  43. Code Migration v1 — evaluation methodology and results ↗

    Vals AI · 2026

  44. CorpFin — evaluation methodology and results ↗

    Vals AI · 2026

  45. Excel Modeling Benchmark v1 — evaluation methodology and results ↗

    Vals AI · 2026

  46. Finance Agent v2 — evaluation methodology and results ↗

    Vals AI · 2026

  47. GPQA Diamond — evaluation methodology and results ↗

    Vals AI · 2026

  48. Harvey's Legal Agent Benchmark v1 — evaluation methodology and results ↗

    Vals AI · 2026

  49. LiveCodeBench — evaluation methodology and results ↗

    Vals AI · 2026

  50. LegalBench — evaluation methodology and results ↗

    Vals AI · 2026

  51. Legal Research Bench — evaluation methodology and results ↗

    Vals AI · 2026

  52. MedQA — evaluation methodology and results ↗

    Vals AI · 2026

  53. MMLU Pro — evaluation methodology and results ↗

    Vals AI · 2026

  54. MMMU Pro — evaluation methodology and results ↗

    Vals AI · 2026

  55. MortgageTax — evaluation methodology and results ↗

    Vals AI · 2026

  56. ProgramBench — evaluation methodology and results ↗

    Vals AI · 2026

  57. ProofBench — evaluation methodology and results ↗

    Vals AI · 2026

  58. Public Benefits Bench — evaluation methodology and results ↗

    Vals AI · 2026

  59. Public Benefits Bench v1 — evaluation methodology and results ↗

    Vals AI · 2026

  60. SAGE — evaluation methodology and results ↗

    Vals AI · 2026

  61. SkillsBench — evaluation methodology and results ↗

    Vals AI · 2026

  62. SWE-bench Verified — evaluation methodology and results ↗

    Vals AI · 2026

  63. Tax Agent Bench — evaluation methodology and results ↗

    Vals AI · 2026

  64. TaxEval — evaluation methodology and results ↗

    Vals AI · 2026

  65. Vals Multimodal Index — evaluation methodology and results ↗

    Vals AI · 2026

  66. Vibe Code Bench — evaluation methodology and results ↗

    Vals AI · 2026

Names used in source tables

Claude Sonnet 4.6 · Claude Sonnet 4.6 (16k thinking) · Claude Sonnet 4.6 (32k thinking) · Claude Sonnet 4.6 (Claude Code) · Claude Sonnet 4.6 (Claude Code) max · Claude Sonnet 4.6 (Max) · Claude Sonnet 4.6 (Mini-SWE-agent) · Claude Sonnet 4.6 (OpenHands) · Claude Sonnet 4.6 (high) · Claude Sonnet 4.6 (low) · Claude Sonnet 4.6 (max) · Claude Sonnet 4.6 (medium) · Claude Sonnet 4.6 (mini-SWE-agent) max · Claude Sonnet 4.6 (no thinking) · Claude Sonnet 4.6 (unknown thinking) · Sonnet 4.6 (Claude Code) · Sonnet-4.6 (Claude Code) · claude-sonnet-4-6