Claude Sonnet 4.6

Leaderboards

Where Claude Sonnet 4.6 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 4.6 has a published result on, out of 100%, with its rank among the models that took it.

• GPQA

  #30 / 256

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

  89.9%

  Verified

• SWE-Bench Verified

  #17 / 120

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

  79.6%

  Self-reported

• Humanity's Last Exam

  #25 / 108

  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

  49.0%

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

• MMMU-Pro

  #32 / 73

  College-level questions that need both an image and text to answer.

  75.6%

  Self-reported · In the vision score

• BrowseComp

  #30 / 68

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

  74.7%

  Self-reported · In the agents score

• Terminal-Bench 2.0

  #23 / 54

  Getting real work done in a command-line terminal.

  59.1%

  Self-reported

• MMMLU

  #12 / 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.

  89.3%

  Self-reported

• DeepSWE 1.1

  #49 / 50

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

  30.0%

  Self-reported · In the coding score

• Multi-Challenge

  #29 / 41

  MultiChallenge is a realistic multi-turn conversation evaluation benchmark that challenges frontier LLMs across four key categories: instruction retention (maintaining instructions throughout conversations), inference memory (recalling and connecting details from previous turns), reliable versioned editing (adapting to evolving instructions during collaborative editing), and self-coherence (avoiding contradictions in responses). The benchmark evaluates models on sustained, contextually complex dialogues across diverse topics including travel planning, technical documentation, and professional communication.

  41.7%

  Verified

• LiveBench

  #13 / 38

  LiveBench is a challenging, contamination-limited LLM benchmark that addresses test set contamination by releasing new questions monthly based on recently-released datasets, arXiv papers, news articles, and IMDb movie synopses. It comprises tasks across math, coding, reasoning, language, instruction following, and data analysis with verifiable, objective ground-truth answers.

  75.5%

  Self-reported · In the text score

• MCP Atlas

  #32 / 38

  MCP Atlas is a benchmark for evaluating AI models on scaled tool use capabilities, measuring how well models can coordinate and utilize multiple tools across complex multi-step tasks.

  61.3%

  Self-reported

• Tau2 Telecom

  #7 / 36

  τ²-Bench telecom domain evaluates conversational agents in a dual-control environment modeled as a Dec-POMDP, where both agent and user use tools in shared telecommunications troubleshooting scenarios that test coordination and communication capabilities.

  97.9%

  Self-reported

• MRCR v2 (8-needle)

  #12 / 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.

  51.2%

  Verified · In the text score

• Finance Agent v2

  #9 / 28

  Finance Agent v2 is an agentic financial-analysis benchmark from Vals that evaluates models on real-world finance workflows, measuring their ability to retrieve and reason over financial documents, perform multi-step calculations, and produce accurate analyses.

  51.0%

  Self-reported

• Tau2 Retail

  #2 / 27

  τ²-bench retail domain evaluates conversational AI agents in customer service scenarios within a dual-control environment where both agent and user can interact with tools. Tests tool-agent-user interaction, rule adherence, and task consistency in retail customer support contexts.

  91.7%

  Self-reported

• HealthBench Hard

  #19 / 24

  A challenging variation of HealthBench that evaluates large language models' performance and safety in healthcare through 5,000 multi-turn conversations with particularly rigorous evaluation criteria validated by 262 physicians from 60 countries

  17.7%

  Verified

• LongBench v2

  #23 / 24

  LongBench v2 is a benchmark designed to assess the ability of LLMs to handle long-context problems requiring deep understanding and reasoning across real-world multitasks. It consists of 503 challenging multiple-choice questions with contexts ranging from 8k to 2M words across six major task categories: single-document QA, multi-document QA, long in-context learning, long-dialogue history understanding, code repository understanding, and long structured data understanding.

  5.6%

  Verified

• OSWorld

  #4 / 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

  72.5%

  Self-reported

• FACTS Grounding

  #15 / 20

  A benchmark evaluating language models' ability to generate factually accurate and well-grounded responses based on long-form input context, comprising 1,719 examples with documents up to 32k tokens requiring detailed responses that are fully grounded in provided documents

  63.6%

  Verified

• ARC-AGI v2

  #7 / 19

  ARC-AGI-2 is an upgraded benchmark for measuring abstract reasoning and problem-solving abilities in AI systems through visual grid transformation tasks. It evaluates fluid intelligence via input-output grid pairs (1x1 to 30x30) using colored cells (0-9), requiring models to identify underlying transformation rules from demonstration examples and apply them to test cases. Designed to be easy for humans but challenging for AI, focusing on core cognitive abilities like spatial reasoning, pattern recognition, and compositional generalization.

  58.3%

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

• Legal Agent Benchmark

  #7 / 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.

  5.4%

  Self-reported

• HealthBench (hard)

  #12 / 13

  17.7%

  Unverified

• GPQA (diamond)

  #8 / 10

  Graduate-level Google-Proof Q&A benchmark with paper-aligned full and diamond subsets.

  80.8%

  Unverified

• mrcr-v2 (8needle_32768_65536)

  #1 / 10

  51.2%

  Unverified

• mrcr-v2 (8needle_4096_8192)

  #4 / 10

  64.8%

  Unverified

• MultiChallenge (default)

  #5 / 9

  41.7%

  Unverified

• mrcr-v2 (8needle_upto_128k)

  #2 / 9

  50.8%

  Unverified

• Finance Agent

  #2 / 8

  Finance Agent is a benchmark for evaluating AI models on agentic financial analysis tasks, testing their ability to process financial data, perform calculations, and generate accurate analyses across various financial domains.

  63.3%

  Self-reported

• facts-grounding (default)

  #6 / 8

  FACTS Grounding 1.0 — 860 public examples from Google DeepMind (CC-BY-4.0)

  63.6%

  Unverified

• LongBench-V2 (short)

  #6 / 6

  5.6%

  Unverified