Qwen3.7 Max

Leaderboards

Where Qwen3.7 Max 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 Qwen3.7 Max has a published result on, out of 100%, with its rank among the models that took it.

• GPQA

  #14 / 256

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

  92.4%

  Self-reported

• MMLU-Pro

  #2 / 142

  Broad knowledge and reasoning across 14 subjects, ten answer choices each.

  89.6%

  Self-reported

• SWE-Bench Verified

  #12 / 120

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

  80.4%

  Self-reported

• Humanity's Last Exam

  #36 / 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

  41.4%

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

• IFEval

  #3 / 70

  Instruction-Following Evaluation (IFEval) benchmark for large language models, focusing on verifiable instructions with 25 types of instructions and around 500 prompts containing one or more verifiable constraints

  94.3%

  Self-reported

• LiveCodeBench v6

  #3 / 67

  LiveCodeBench is a holistic and contamination-free evaluation benchmark for large language models for code. It continuously collects new problems from programming contests (LeetCode, AtCoder, CodeForces) and evaluates four different scenarios: code generation, self-repair, code execution, and test output prediction. Problems are annotated with release dates to enable evaluation on unseen problems released after a model's training cutoff.

  91.6%

  Self-reported

• SWE-Bench Pro

  #20 / 61

  Harder, longer software engineering tasks from real codebases.

  60.6%

  Self-reported · In the coding score

• Terminal-Bench 2.0

  #9 / 54

  Getting real work done in a command-line terminal.

  69.7%

  Self-reported

• MMMLU

  #8 / 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.

  90.3%

  Self-reported

• MMLU-Redux

  #1 / 50

  An improved version of the MMLU benchmark featuring manually re-annotated questions to identify and correct errors in the original dataset. Provides more reliable evaluation metrics for language models by addressing dataset quality issues found in the original MMLU.

  95.0%

  Self-reported

• SWE-bench Multilingual

  #10 / 49

  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

• IFBench

  #11 / 44

  Instruction Following Benchmark evaluating model's ability to follow complex instructions

  79.1%

  Self-reported

• LiveBench

  #19 / 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.

  74.3%

  Self-reported · In the text score

• MCP Atlas

  #13 / 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.

  76.4%

  Self-reported

• SuperGPQA

  #1 / 37

  SuperGPQA is a comprehensive benchmark that evaluates large language models across 285 graduate-level academic disciplines. The benchmark contains 25,957 questions covering 13 broad disciplinary areas including Engineering, Medicine, Science, and Law, with specialized fields in light industry, agriculture, and service-oriented domains. It employs a Human-LLM collaborative filtering mechanism with over 80 expert annotators to create challenging questions that assess graduate-level knowledge and reasoning capabilities.

  73.6%

  Self-reported

• MMLU-ProX

  #1 / 32

  Extended version of MMLU-Pro providing additional challenging multiple-choice questions for evaluating language models across diverse academic and professional domains. Built on the foundation of the Massive Multitask Language Understanding benchmark framework.

  87.0%

  Self-reported

• Include

  #2 / 31

  Include benchmark - specific documentation not found in official sources

  86.2%

  Self-reported

• SciCode

  #6 / 30

  SciCode is a research coding benchmark curated by scientists that challenges language models to code solutions for scientific problems. It contains 338 subproblems decomposed from 80 challenging main problems across 16 natural science sub-fields including mathematics, physics, chemistry, biology, and materials science. Problems require knowledge recall, reasoning, and code synthesis skills.

  53.5%

  Self-reported

• Finance Agent v2

  #10 / 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.

  48.4%

  Self-reported

• NL2Repo

  #14 / 25

  NL2Repo evaluates long-horizon coding capabilities including repository-level understanding, where models must generate or modify code across entire repositories from natural language specifications.

  47.2%

  Self-reported · In the coding score

• WMT24++

  #3 / 24

  WMT24++ is a comprehensive multilingual machine translation benchmark that expands the WMT24 dataset to cover 55 languages and dialects. It includes human-written references and post-edits across four domains (literary, news, social, and speech) to evaluate machine translation systems and large language models across diverse linguistic contexts.

  85.8%

  Self-reported

• PolyMATH

  #1 / 23

  Polymath is a challenging multi-modal mathematical reasoning benchmark designed to evaluate the general cognitive reasoning abilities of Multi-modal Large Language Models (MLLMs). The benchmark comprises 5,000 manually collected high-quality images of cognitive textual and visual challenges across 10 distinct categories, including pattern recognition, spatial reasoning, and relative reasoning.

  86.5%

  Self-reported

• IMO-AnswerBench

  #3 / 22

  IMO-AnswerBench is a benchmark for evaluating mathematical reasoning capabilities on International Mathematical Olympiad (IMO) problems, focusing on answer generation and verification.

  90.0%

  Self-reported

• BFCL-V4

  #1 / 21

  Berkeley Function Calling Leaderboard V4 (BFCL-V4) evaluates LLMs on their ability to accurately call functions and APIs, including simple, multiple, parallel, and nested function calls across diverse programming scenarios.

  75.0%

  Self-reported

• Global PIQA

  #3 / 15

  Global PIQA is a multilingual commonsense reasoning benchmark that evaluates physical interaction knowledge across 100 languages and cultures. It tests AI systems' understanding of physical world knowledge in diverse cultural contexts through multiple choice questions about everyday situations requiring physical commonsense.

  91.4%

  Self-reported

• Claw-Eval

  #5 / 14

  Claw-Eval tests real-world agentic task completion across complex multi-step scenarios, evaluating a model's ability to use tools, navigate environments, and complete end-to-end tasks autonomously.

  65.2%

  Self-reported

• HMMT Feb 26

  #1 / 13

  HMMT February 2026 is a math competition benchmark based on problems from the Harvard-MIT Mathematics Tournament, testing advanced mathematical problem-solving and reasoning.

  97.1%

  Self-reported

• MAXIFE

  #1 / 12

  MAXIFE is a multilingual benchmark evaluating LLMs on instruction following and execution across multiple languages and cultural contexts.

  89.2%

  Self-reported

• NOVA-63

  #2 / 12

  NOVA-63 is a multilingual evaluation benchmark covering 63 languages, designed to assess LLM performance across diverse linguistic contexts and tasks.

  59.0%

  Self-reported

• MathArena Apex

  #6 / 11

  MathArena Apex is a challenging math contest benchmark featuring the most difficult mathematical problems designed to test advanced reasoning and problem-solving abilities of AI models. It focuses on olympiad-level mathematics and complex multi-step mathematical reasoning.

  44.5%

  Self-reported

• VITA-Bench

  #2 / 11

  VITA-Bench evaluates AI agents on real-world virtual task automation, measuring their ability to complete complex multi-step tasks in simulated environments.

  47.9%

  Self-reported

• SkillsBench

  #3 / 10

  SkillsBench evaluates coding agents on self-contained programming tasks, measuring practical engineering skills across diverse software development scenarios.

  59.2%

  Self-reported

• MCP-Mark

  #2 / 9

  MCP-Mark evaluates LLMs on their ability to use Model Context Protocol (MCP) tools effectively, testing tool discovery, selection, invocation, and result interpretation across diverse MCP server scenarios.

  60.8%

  Self-reported

• CoWorkBench

  #5 / 6

  CoWorkBench is Qwen's internal cowork benchmark for evaluating long-horizon office and productivity agent tasks across domains such as computer science, finance, law, and medicine.

  67.2%

  Self-reported

• CritPT

  #3 / 6

  CritPT is a challenging reasoning benchmark reported by Qwen for evaluating frontier mathematical and critical problem-solving capability.

  11.4%

  Self-reported

• SpreadSheetBench-v1

  #2 / 4

  SpreadSheetBench-v1 evaluates office automation agents on spreadsheet reasoning and manipulation tasks, measuring the ability to analyze, transform, and operate on spreadsheet data through tools.

  87.0%

  Self-reported

• ZClawBench

  #1 / 4

  ZClawBench evaluates Claw-style agent task execution quality, measuring a model's ability to autonomously complete complex multi-step coding tasks in real-world environments.

  64.3%

  Self-reported

• MRCR 128K (8-needle)

  #1 / 2

  MRCR (Multi-Round Coreference Resolution) at 128K context length with 8 needles. Models must navigate long conversations to reproduce specific model outputs, testing attention and reasoning across 128K-token contexts with 8 items to retrieve.

  90.4%

  Unverified

• QwenSVG

  #2 / 2

  QwenSVG is Qwen's internal SVG generation benchmark for evaluating front-end and visual code generation. Scores are reported as BT/Elo ratings from auto-rendered outputs judged by a multimodal evaluator.

  80.4%

  Unverified

• QwenWebBench

  #1 / 2

  QwenWebBench is an internal front-end code generation benchmark by Qwen. It is bilingual (EN/CN) and spans 7 categories (Web Design, Web Apps, Games, SVG, Data Visualization, Animation, and 3D), using auto-render plus a multimodal judge for code and visual correctness. Scores are reported as BT/Elo ratings.

  78.4%

  Unverified

• QwenWorldBench

  #2 / 2

  QwenWorldBench is Qwen's internal benchmark for evaluating LLMs as world models that simulate agentic environments across Terminal, SWE, MCP, Search, OS, Android, and Web domains.

  57.3%

  Unverified

• Kernel Bench L3

  #1 / 1

  Kernel Bench L3 evaluates agentic GPU kernel optimization across 50 problems. Qwen reports two metrics for this benchmark: median per-problem speedup over the PyTorch eager reference and the fraction of problems faster than torch.compile.

  96.0%

  Unverified