Muse Wonder

Leaderboards

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

• DeepSWE 1.1

  #3 / 50

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

  78.8%

  Self-reported · In the coding score

• Terminal-Bench 4.0

  #8 / 28

  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.

  52.0%

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

• HealthBench Professional

  #9 / 19

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

  60.5%

  Self-reported

• Program Bench

  #10 / 18

  Program Bench evaluates code-generation agents by asking them to recreate a program's behavior from only a compiled binary and documentation. It spans 200 tasks from small CLI tools to large systems such as FFmpeg and SQLite, with submissions judged against more than 248,000 fuzz-generated behavioral tests.

  50.0%

  Self-reported · In the coding score

• GDP.pdf

  #7 / 14

  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.

  30.8%

  Self-reported

• Artificial Analysis

  #8 / 13

  Artificial Analysis benchmark evaluates AI models across quality, speed, and pricing dimensions, providing a composite assessment of model capabilities for real-world usage.

  52.0%

  Self-reported

• SWE-Marathon

  #3 / 10

  SWE-Marathon is an ultra-long-horizon software engineering benchmark covering tasks such as building compilers, optimizing kernels, and developing production-grade services. It measures whether agents can sustain quality across extremely long engineering trajectories.

  53.1%

  Self-reported

• JobBench

  #3 / 6

  JobBench evaluates agents on work-related tasks, as reported in the MiMo-V2.6 release.

  70.5%

  Self-reported

• OSWorld 2.1 (partial)

  #4 / 6

  Partial-credit computer-use score on OSWorld 2.1, as reported in the Claude Sonnet 5.5 release. Separate from OSWorld 2.0 and strict completion scores.

  72.6%

  Self-reported

• OSWorld 2.1 (strict)

  #5 / 6

  Strict pass rate on OSWorld 2.1: fraction of tasks where every graded checkpoint is satisfied. Separate from the OSWorld 2.1 partial-credit score and from OSWorld 2.0.

  34.3%

  Self-reported

• Vals Finance Agent v2

  #5 / 6

  Vals Finance Agent v2 evaluates AI agents on realistic financial analyst tasks.

  56.6%

  Self-reported

• AutomationBench-AA

  #4 / 5

  AutomationBench-AA is Artificial Analysis's independently run version of AutomationBench, covering 657 real-world SaaS workflow tasks across 40 simulated applications (e.g. Gmail, Slack, Salesforce, HubSpot). It scores the share of objectives an agent completes without violating business guardrails, using a private held-out task set.

  39.1%

  Self-reported

• Harvey LAB-AA

  #4 / 5

  Harvey LAB-AA evaluates model performance on complex legal workflows.

  10.0%

  Self-reported

• FocusBench

  #3 / 4

  Internal personal agent evaluation across housing, travel, IT, and finance; persistence through blockers and adaptation across interactions and changing environments.

  44.9%

  Self-reported

• MemoryBench

  #1 / 4

  Internal recall evaluation over 64K-100K-token conversations containing distractors, conflicting facts, and later updates; LLM judge with weighted question-specific rubrics.

  85.6%

  Self-reported

• Personal Superintelligence Index

  #3 / 4

  Internal composite of Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT, and DRACO.

  54.3%

  Self-reported

• Terminal-Bench 4.0-AA

  #3 / 4

  Artificial Analysis reported accuracy on terminal tasks with executable final-container-state verifiers.

  52.0%

  Self-reported

• AA Intelligence Index v4.3

  #2 / 3

  Version 4.3 of the Artificial Analysis Intelligence Index. Scores are stored as fractions of 100.

  52.0%

  Self-reported