Claude Sonnet 4.5

Leaderboards

Where Claude Sonnet 4.5 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 4.5 has a published result on, out of 100%, with its rank among the models that took it.

• GPQA

  #81 / 256

  Graduate-level science questions written to be hard to look up.

  83.4%

  Self-reported

• AIME 2025

  #59 / 123

  Competition maths problems from the 2025 American Invitational Mathematics Exam.

  87.0%

  Self-reported

• MMMLU

  #13 / 51

  Multilingual Massive Multitask Language Understanding dataset released by OpenAI, featuring professionally translated MMLU test questions across 14 languages including Arabic, Bengali, German, Spanish, French, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Swahili, Yoruba, and Chinese. Contains approximately 15,908 multiple-choice questions per language covering 57 subjects.

  89.1%

  Self-reported

• MRCR v2 (8-needle)

  #13 / 32

  MRCR v2 (8-needle) is a variant of the Multi-Round Coreference Resolution benchmark that includes 8 needle items to retrieve from long contexts. This tests models' ability to simultaneously track and reason about multiple pieces of information across extended conversations.

  47.5%

  Verified · In the text score

• TAU-bench Retail

  #1 / 25

  A benchmark for evaluating tool-agent-user interaction in retail environments. Tests language agents' ability to handle dynamic conversations with users while using domain-specific API tools and following policy guidelines. Evaluates agents on tasks like order cancellations, address changes, and order status checks through multi-turn conversations.

  86.2%

  Self-reported

• Terminal-Bench

  #1 / 25

  Terminal-Bench is a benchmark for testing AI agents in real terminal environments. It evaluates how well agents can handle real-world, end-to-end tasks autonomously, including compiling code, training models, setting up servers, system administration, security tasks, data science workflows, and cybersecurity vulnerabilities. The benchmark consists of a dataset of ~100 hand-crafted, human-verified tasks and an execution harness that connects language models to a terminal sandbox.

  50.0%

  Self-reported

• TAU-bench Airline

  #1 / 23

  Part of τ-bench (TAU-bench), a benchmark for Tool-Agent-User interaction in real-world domains. The airline domain evaluates language agents' ability to interact with users through dynamic conversations while following domain-specific rules and using API tools. Agents must handle airline-related tasks and policies reliably.

  70.0%

  Self-reported

• OSWorld

  #9 / 21

  OSWorld: The first-of-its-kind scalable, real computer environment for multimodal agents, supporting task setup, execution-based evaluation, and interactive learning across Ubuntu, Windows, and macOS with 369 computer tasks involving real web and desktop applications, OS file I/O, and multi-application workflows

  61.4%

  Self-reported

• mrcr-v2 (8needle_4096_8192)

  #6 / 10

  47.5%

  Unverified

• MMMUval

  #3 / 4

  Validation set for MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning) benchmark, designed to evaluate multimodal models on massive multi-discipline tasks demanding college-level subject knowledge and deliberate reasoning across Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering.

  77.8%

  Self-reported

• SWE-bench Verified (Agentic Coding)

  #1 / 2

  SWE-bench Verified is a human-filtered subset of 500 software engineering problems drawn from real GitHub issues across 12 popular Python repositories. Given a codebase and an issue description, language models are tasked with generating patches that resolve the described problems. This benchmark evaluates AI's real-world agentic coding skills by requiring models to navigate complex codebases, understand software engineering problems, and coordinate changes across multiple functions, classes, and files to fix well-defined issues with clear descriptions.

  77.2%

  Unverified