Share promptsSign inCreate team library
The Chat Judge verdicts

Which AI is best at what?

Public benchmark results for 19 tests across 9 fields, charted side by side. Every score links to the leaderboard it came from.

Showing Meta's best-ranked model on each benchmark. Try prompts for Muse →

1 benchmarkUpdated 19 Oct 2026Overall workOne number for agentic work across finance, coding, law and tax.
  • Vals Index Wonder Full #4 · 66.79%
See charts 3 benchmarksUpdated 19 Oct 2026LegalDrafting, extraction, legal reasoning and US legal research.
  • Legal Benchmarks leaderboard Muse Spark 1.3 #4 · 87.0%
  • LegalBench Wonder Full #4 · 87.42%
  • Legal Research Bench Muse Spark 1.3 Max #1 · 55.29%
See charts 3 benchmarksUpdated 21 Oct 2026Health & medicalPhysician-graded health conversations, medical coding and exam questions.
  • MedCode Muse Spark #16 · 51.31%
See charts 2 benchmarksUpdated 19 Oct 2026FinanceAnalyst research on SEC filings and long credit agreements.
  • Finance Agent Wonder Full #3 · 67.58%
  • CorpFin v2 Muse Spark 1.1 #4 · 71.29%
See charts 2 benchmarksUpdated 1 Sep 2026CodingFixing real GitHub issues and solving fresh programming problems.
  • SWE-bench Verified Muse Spark 1.2 #14 · 86.6%
See charts 2 benchmarksUpdated 19 Oct 2026Agents & tool useMulti-step tasks an agent completes alone in a terminal.
  • Terminal-Bench 4.0 Wonder Full #5 · 58.08%
  • Terminal-Bench 2.1 Muse Spark 1.3 Max #12 · 79.03%
See charts 4 benchmarksUpdated 21 Oct 2026Reasoning & knowledgeExpert exams, graduate science, contest maths and broad knowledge.
  • Humanity's Last Exam Muse Spark #7 · 40.56%
  • GPQA Diamond Muse Spark 1.1 #23 · 91.16%
  • AIME Muse Spark #3 · 96.88%
  • MMLU Pro Muse Spark 1.1 #17 · 88.73%
See charts 1 benchmarkUpdated 20 Oct 2026Chat preferenceWhich answers real people prefer in blind side-by-side votes.
  • LMArena Text Arena Muse Wonder (Max) #2 · 1510
See charts 1 benchmarkUpdated 9 Jan 2026MultilingualReasoning in languages other than English.
  • MGSM Llama 4 Maverick #18 · 92.44%
See charts

ChatJudge does not run these tests. Scores are copied as published by each source, with a link and the date we read them, and only change when we refresh from the source. Different leaderboards use different settings, so compare models within one chart, not across charts. Spotted a mistake? Tell us.