Claude Sonnet 5

Leaderboards

Where Claude Sonnet 5 sits among the top 20 in each category.

By task

Scored from votes on each kind of prompt alone, on the same 1,000-point scale.

Test results

Every test Claude Sonnet 5 has a published result on, out of 100%, with its rank among the models that took it.

• SWE-Bench Verified

  #6 / 123

  Fixing real bugs in real GitHub projects, checked by the projects' own tests.

  85.2%

  Self-reported

• Humanity's Last Exam

  #10 / 115

  Humanity's Last Exam (HLE) is a multi-modal academic benchmark with 2,500 questions across mathematics, humanities, and natural sciences, designed to test LLM capabilities at the frontier of human knowledge with unambiguous, verifiable solutions

  57.4%

  Self-reported · In the text score· In the vision score

• BrowseComp

  #13 / 73

  Finding hard-to-locate facts by browsing the web.

  84.7%

  Self-reported · In the agents score

• SWE-Bench Pro

  #14 / 63

  Harder, longer software engineering tasks from real codebases.

  63.2%

  Self-reported · In the coding score

• CharXiv-R

  #13 / 61

  Reading and reasoning about charts from scientific papers.

  88.3%

  Self-reported

• Terminal-Bench 2.0

  #3 / 54

  Getting real work done in a command-line terminal.

  80.4%

  Self-reported

• DeepSWE 1.1

  #40 / 52

  DeepSWE 1.1 evaluates software engineering agents using the mini-swe-agent harness.

  54.0%

  Self-reported · In the coding score

• SWE-bench Multilingual

  #10 / 50

  A multilingual benchmark for issue resolving in software engineering that covers Java, TypeScript, JavaScript, Go, Rust, C, and C++. Contains 1,632 high-quality instances carefully annotated from 2,456 candidates by 68 expert annotators, designed to evaluate Large Language Models across diverse software ecosystems beyond Python.

  78.3%

  Self-reported

• Toolathlon

  #17 / 42

  Tool Decathlon is a comprehensive benchmark for evaluating AI agents' ability to use multiple tools across diverse task categories. It measures proficiency in tool selection, sequencing, and execution across ten different tool-use scenarios.

  54.3%

  Self-reported · In the agents score

• Terminal-Bench 4.0

  #29 / 31

  Terminal-Bench 4.0 evaluates AI agents on real-world work in terminal and command-line containerized environments across 66 tasks, with emphasis on science-adjacent and frontier engineering problems. Relative to earlier Terminal-Bench releases, 4.0 increases timeouts and adaptively raises RAM/CPU on selected tasks to reduce harness and resource confounds.

  12.4%

  Self-reported · In the coding score· In the agents score

• OSWorld-Verified

  #7 / 28

  OSWorld-Verified is a verified subset of OSWorld, a scalable real computer environment for multimodal agents supporting task setup, execution-based evaluation, and interactive learning across Ubuntu, Windows, and macOS.

  81.2%

  Self-reported

• AutomationBench

  #27 / 27

  AutomationBench is a tool-use benchmark that evaluates AI agents on automating real-world workflows, testing their ability to orchestrate tools and complete multi-step automation tasks.

  13.5%

  Self-reported · In the agents score

• FrontierCode 1.1

  #14 / 24

  FrontierCode 1.1 evaluates whether coding-agent changes are mergeable, using unit tests, maintainer-defined rubrics, and verifiers. Runs flagged for unfair internet use receive a zero score.

  42.7%

  Self-reported · In the coding score

• HealthBench Professional

  #14 / 20

  HealthBench Professional evaluates model capability and safety for clinician use cases using real clinician-style chats and physician-authored grading rubrics.

  57.8%

  Self-reported

• GDP.pdf

  #1 / 15

  GDP.pdf is a knowledge-work vision benchmark that evaluates models on economically valuable professional tasks presented as visual documents (PDFs), testing document-based reasoning, chart and table interpretation, and problem solving without tools.

  81.6%

  Self-reported

• Legal Agent Benchmark

  #6 / 15

  The Legal Agent Benchmark (LAB) is Harvey's open-source benchmark for evaluating AI agents on complex, long-horizon legal work. Tasks are scored under an all-pass standard against expert-curated rubrics, where a task passes only if every required rubric criterion (facts, conclusions, citations, structure, and analytical moves) passes.

  5.8%

  Self-reported

• OfficeQA Pro

  #11 / 13

  OfficeQA Pro evaluates AI models on professional knowledge-work questions and tasks drawn from real office workflows, including document analysis, spreadsheet reasoning, and information synthesis across business domains.

  59.4%

  Self-reported

• BenchCAD

  #8 / 8

  BenchCAD is a benchmark for programmatic CAD reasoning built from 17,900 execution-verified CadQuery programs spanning 106 industrial part families, roughly half anchored to real ISO, DIN, EN, ASME, and IEC specification tables. It decomposes CAD capability into matched tasks; the Vision2Code task requires models to generate CadQuery code from multi-view renders, scored by voxel IoU against the reference geometry.

  37.3%

  Self-reported

• SWE-Bench Multimodal

  #7 / 7

  SWE-Bench Multimodal extends SWE-Bench to evaluate language models on software engineering tasks that involve visual inputs such as screenshots, UI mockups, and diagrams alongside code understanding.

  28.1%

  Self-reported

• ArXivMath

  #3 / 5

  ArXivMath is a final-answer benchmark of research-level mathematics maintained by MathArena. Problems are extracted monthly from recent arXiv paper abstracts, then filtered through automated and manual checks to ensure they are self-contained, non-trivial, and verifiable. Because problems are drawn from active research, the benchmark is more realistic and more closely connected to mathematical research than contest or olympiad benchmarks.

  72.2%

  Self-reported

• FrontierCode

  #4 / 4

  FrontierCode is Cognition's coding evaluation that tests whether models can pass difficult coding tasks while meeting the standards of high-quality production codebases. The Diamond subset contains the hardest problems.

  38.8%

  Self-reported

• USAMO 2026

  #2 / 3

  USAMO 2026 evaluates models on the six problems from the 2026 United States of America Mathematical Olympiad, requiring rigorous proof-based reasoning. Following the MathArena methodology, proofs are graded against human expert rubrics by dual strong judge models, with the minimum of the two scores taken as final. Maximum score is 42 points.

  79.5%

  Self-reported

• ChartMuseum

  #1 / 1

  ChartMuseum is a chart question-answering benchmark of 1,162 expert-annotated questions over real-world chart images drawn from 184 sources, including academic figures, infographics, and unconventional chart designs. It specifically targets questions that require visual reasoning, such as comparing unlabeled visual elements, tracking trajectories, and judging spatial relationships.

  86.7%

  Unverified