RLTasksby ulam.ai
Models / Anthropic

Anthropic / MODEL EVIDENCE

Claude Sonnet 4.5

55 benchmarks · 105 observations · 62 sources · released 2025-09-29

Scores at a glance

55 benchmarks · large dot: Claude Sonnet 4.5; small dots: closest models where reported

GPQA Diamond ↗82.32%
82.3282.32

Claude Sonnet 4.5 (59k thinking) · GPQA diamond · Best score (across scorers) · % ↗

4 more reported results

81.73% · Claude Sonnet 4.5 (32k thinking) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

78.79% · Claude Sonnet 4.5 (16k thinking) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

73.74% · Claude Sonnet 4.5 (no thinking) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

81.63 % · Claude Sonnet 4.5 (Thinking) · GPQA Diamond · Accuracy · Vals AI · GPQA Diamond ↗

FrontierMath Tiers 1-3 (v1, old) ↗15.22%
15.2215.22

Claude Sonnet 4.5 (32k thinking) · FrontierMath-2025-02-28-Private · Best score (across scorers) · % ↗

2 more reported results

13.49% · Claude Sonnet 4.5 (59k thinking) · FrontierMath-2025-02-28-Private · Best score (across scorers) · % · Epoch AI · FrontierMath-2025-02-28-Private ↗

9.31% · Claude Sonnet 4.5 (no thinking) · FrontierMath-2025-02-28-Private · Best score (across scorers) · % · Epoch AI · FrontierMath-2025-02-28-Private ↗

MATH Level 5 ↗97.73%
3.2998.13

Claude Sonnet 4.5 (32k thinking) · MATH level 5 · Best score (across scorers) · % ↗

OTIS Mock AIME 2024-2025 ↗77.78%
77.7877.78

Claude Sonnet 4.5 (59k thinking) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % ↗

3 more reported results

77.78% · Claude Sonnet 4.5 (32k thinking) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗

71.11% · Claude Sonnet 4.5 (16k thinking) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗

35.56% · Claude Sonnet 4.5 (no thinking) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗

Cybench ↗60%
6082

Claude Sonnet 4.5 (unknown thinking) · Cybench · Unguided % Solved · % ↗

1 more reported result

55% · Claude Sonnet 4.5 (no thinking) · Cybench · Unguided % Solved · % · Epoch AI · Cybench ↗

GSO ↗14.71%
1.344.12

Claude Sonnet 4.5 (unknown thinking) · GSO-Bench · Score OPT@1 · % ↗

1 more reported result

14.7% · Claude Sonnet 4.5 (no thinking) · GSO-Bench · Score OPT@1 · % · Epoch AI · GSO-Bench ↗

SimpleBench ↗54.3%
54.354.3

Claude Sonnet 4.5 (12k thinking) · SimpleBench · Score (AVG@5) · % ↗

1 more reported result

54.3% · Claude Sonnet 4.5 (unknown thinking) · SimpleBench · Score (AVG@5) · % · Epoch AI · SimpleBench ↗

Terminal-Bench 2.0 ↗46.52%
3.0784.72

Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % ↗

11 more reported results

43.15% · Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.81% · Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.7% · Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.58% · Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.53% · Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

40.06% · Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.8% · Claude Sonnet 4.5 (no thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.7% · Claude Sonnet 4.5 (unknown thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.6% · Claude Sonnet 4.5 (no thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

42.5% · Claude Sonnet 4.5 (no thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

40.1% · Claude Sonnet 4.5 (no thinking) · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

VPCT ↗39.8%
39.839.8

Claude Sonnet 4.5 (32k thinking) · VPCT · Correct · % ↗

1 more reported result

38% · Claude Sonnet 4.5 (no thinking) · VPCT · Correct · % · Epoch AI · VPCT ↗

WeirdML (v2) ↗47.71%
47.7147.71

Claude Sonnet 4.5 (16k thinking) · WeirdML · Accuracy · % ↗

1 more reported result

46.7% · Claude Sonnet 4.5 (no thinking) · WeirdML · Accuracy · % · Epoch AI · WeirdML ↗

CritPt ↗1.14%
032.29

Claude Sonnet 4.5 (unknown thinking) · CritPt · Accuracy · % ↗

GDPval ↗42.5%
9.949.7

Claude Sonnet 4.5 (no thinking) · GDPval · Win Rate (%) · % ↗

ProofBench ↗19%
0100

Claude Sonnet 4.5 (unknown thinking) · ProofBench · Accuracy · % ↗

LMCA ↗38.76%
2.7868.26

Claude Sonnet 4.5 (unknown thinking) · LMCA · Score · % ↗

DTBench ↗83.2%
41.6598.93

Claude Sonnet 4.5 (unknown thinking) · DTBench · Accuracy · % ↗

SciCode ↗45.72%
1.2766.9

Claude Sonnet 4.5 (unknown thinking) · SciCode · Score · % ↗

EnigmaEval ↗6%
0.3839.28

Claude Sonnet 4.5 (unknown thinking) · EnigmaEval · Accuracy · % ↗

LegalBench ↗84.08 %
32.9788.56

Claude Sonnet 4.5 (Thinking) · LegalBench · Accuracy ↗

CaseLaw v2 ↗62.16 %
43.8479.31

Claude Sonnet 4.5 (Thinking) · CaseLaw · Accuracy ↗

CorpFin v2 ↗61.97 %
61.9761.97

Claude Sonnet 4.5 (Thinking) · CorpFin · Accuracy ↗

1 more reported result

60.8 % · Claude Sonnet 4.5 (Nonthinking) · CorpFin · Accuracy · Vals AI · CorpFin ↗

MedCode ↗44.13 %
44.1344.13

Claude Sonnet 4.5 (Thinking) · MedCode · Accuracy ↗

1 more reported result

40.57 % · Claude Sonnet 4.5 (Nonthinking) · MedCode · Accuracy · Vals AI · MedCode ↗

MedScribe ↗84.52 %
84.5284.52

Claude Sonnet 4.5 (Nonthinking) · MedScribe · Accuracy ↗

1 more reported result

84.1 % · Claude Sonnet 4.5 (Thinking) · MedScribe · Accuracy · Vals AI · MedScribe ↗

MMLU-Pro ↗87.36 %
30.2892.38

Claude Sonnet 4.5 (Thinking) · MMLU Pro · Accuracy ↗

MMMU-Pro ↗79.31 %
22.7790.64

Claude Sonnet 4.5 (Thinking) · MMMU Pro · Accuracy ↗

Vibe Code Bench ↗22.62 %
092.39

Claude Sonnet 4.5 (Thinking) (OpenHands) · Vibe Code Bench · original task set ↗

MCP Atlas ↗59.50%
40.288.1

claude-sonnet-4-5 (thinking) · Scale Labs · Score ↗

EconBench ↗12.9827 error points
12.9812.98

claude-sonnet-4-5 · EconBench · EconBench prototype · risk-consistency error · published prototype observations ↗

NYU CTF Bench ↗27.2%
11.179

Claude Sonnet 4.5 · NYU CTF Bench · Cotool 81-challenge subset · accuracy ↗

AlgoTune ↗1.52 ×
1.312.05

Claude Sonnet 4.5 (high) · Harmonic-mean speedup ↗

Closest models by score

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

o3 →OpenAI
13.6 percentile-point gap

12 of 18 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. Arena leaderboard — WebDev evaluation ↗

    Arena

  2. SciCode Benchmark | Artificial Analysis ↗

    Artificial Analysis

  3. 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

  4. System Card: Claude Opus 4.5 ↗

    Anthropic · 2025

  5. DTBench — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  6. LMCA — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  7. Cybench: A Framework for Evaluating Cybersecurity Capabilities and Risks of Language Models ↗

    Zhang, Andy K., Perry, Neil, Dulepet, Riya, Ji, Joey, Menders, Celeste, Lin, Justin W. et al. · 2024

    All 27 authors

    Zhang, Andy K., Perry, Neil, Dulepet, Riya, Ji, Joey, Menders, Celeste, Lin, Justin W., Jones, Eliot, Hussein, Gashon, Liu, Samantha, Jasper, Donovan, Peetathawatchai, Pura, Glenn, Ari, Sivashankar, Vikram, Zamoshchin, Daniel, Glikbarg, Leo, Askaryar, Derek, Yang, Mike, Zhang, Teddy, Alluri, Rishi, Tran, Nathan, Sangpisit, Rinnara, Yiorkadjis, Polycarpos, Osele, Kenny, Raghupathi, Gautham, Boneh, Dan, Ho, Daniel E., Liang, Percy

  8. FutureSearch Evals ↗

    FutureSearch

  9. EconBench: Economic choice by language models — claude-sonnet-4-5 prototype metrics ↗

    EconBench project · 2026

  10. Algotune — methodology and published results ↗

    Epoch AI · 2026

  11. ARC-AGI-1 ↗

    Epoch AI

  12. ARC-AGI-2 ↗

    Epoch AI

  13. Chess Puzzles ↗

    Epoch AI

  14. CritPt v1 ↗

    Epoch AI

  15. Earthborne Rangers (EBR-bench) ↗

    Epoch AI

  16. FrontierMath Tier 4 (v1, old) ↗

    Epoch AI

  17. FrontierMath Tier 4 (v2) ↗

    Epoch AI

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

    Epoch AI

  19. FrontierMath Tiers 1-3 (v2) ↗

    Epoch AI

  20. GDPval ↗

    Epoch AI

  21. GPQA Diamond ↗

    Epoch AI

  22. Humanity's Last Exam ↗

    Epoch AI

  23. MATH Level 5 ↗

    Epoch AI

  24. Mystery Game Puzzles ↗

    Epoch AI

  25. OTIS Mock AIME 2024-2025 ↗

    Epoch AI

  26. Remote Labor Index ↗

    Epoch AI

  27. SimpleQA Verified ↗

    Epoch AI

  28. SimpleBench ↗

    Epoch AI

  29. SWE-bench Verified ↗

    Epoch AI

  30. Terminal-Bench 2.0 ↗

    Epoch AI

  31. Capabilities & benchmarking — retained external evaluation records ↗

    Epoch AI · 2026

  32. Vending Bench 2 — methodology and published results ↗

    Epoch AI · 2026

  33. Text Arena (Coding) ↗

    Epoch AI

  34. AI benchmarks & capabilities ↗

    Epoch AI

  35. FutureSearch Evals ↗

    FutureSearch

  36. GSO: Challenging Software Optimization Tasks for Evaluating SWE-Agents ↗

    Shetty, Manish, Jain, Naman, Liu, Jinjian, Kethanaboyina, Vijay, Sen, Koushik, Stoica, Ion · 2025

  37. WeirdML ↗

    Håvard Tveit Ihle

  38. EnigmaEval ↗

    Scale Labs / SEAL

  39. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  40. Measuring AI Ability to Complete Long Software Tasks ↗

    METR

  41. 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

  42. NYU CTF Bench — Cotool evaluation, February 2026 ↗

    Cotool · 2026

  43. Terminal-Bench 2.0 leaderboard ↗

    Terminal-Bench Team

  44. AIME — evaluation methodology and results ↗

    Vals AI · 2026

  45. CaseLaw — evaluation methodology and results ↗

    Vals AI · 2026

  46. CorpFin — evaluation methodology and results ↗

    Vals AI · 2026

  47. GPQA Diamond — evaluation methodology and results ↗

    Vals AI · 2026

  48. LiveCodeBench — evaluation methodology and results ↗

    Vals AI · 2026

  49. LegalBench — evaluation methodology and results ↗

    Vals AI · 2026

  50. MedCode — evaluation methodology and results ↗

    Vals AI · 2026

  51. MedQA — evaluation methodology and results ↗

    Vals AI · 2026

  52. MedScribe — evaluation methodology and results ↗

    Vals AI · 2026

  53. MGSM — evaluation methodology and results ↗

    Vals AI · 2026

  54. MMLU Pro — evaluation methodology and results ↗

    Vals AI · 2026

  55. MMMU Pro — evaluation methodology and results ↗

    Vals AI · 2026

  56. MortgageTax — evaluation methodology and results ↗

    Vals AI · 2026

  57. Poker Agent — evaluation methodology and results ↗

    Vals AI · 2026

  58. ProofBench — evaluation methodology and results ↗

    Vals AI · 2026

  59. SAGE — evaluation methodology and results ↗

    Vals AI · 2026

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

    Vals AI · 2026

  61. TaxEval — evaluation methodology and results ↗

    Vals AI · 2026

  62. Vibe Code Bench — evaluation methodology and results ↗

    Vals AI · 2026

Names used in source tables

Claude Sonnet 4.5 · Claude Sonnet 4.5 (12k thinking) · Claude Sonnet 4.5 (16k thinking) · Claude Sonnet 4.5 (1k thinking) · Claude Sonnet 4.5 (2k thinking) · Claude Sonnet 4.5 (32k thinking) · Claude Sonnet 4.5 (48k thinking) · Claude Sonnet 4.5 (59k thinking) · Claude Sonnet 4.5 (8k thinking) · Claude Sonnet 4.5 (Nonthinking) · Claude Sonnet 4.5 (Thinking) · Claude Sonnet 4.5 (Thinking) (Mini-SWE-agent) · Claude Sonnet 4.5 (Thinking) (OpenHands) · Claude Sonnet 4.5 (high) · Claude Sonnet 4.5 (low) · Claude Sonnet 4.5 (no thinking) · Claude Sonnet 4.5 (unknown thinking) · claude-sonnet-4-5 · claude-sonnet-4-5 (thinking)