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 OpenAI's best-ranked model on each benchmark. Try prompts for ChatGPT →

1 benchmarkUpdated 19 Oct 2026Overall workOne number for agentic work across finance, coding, law and tax.
  • Vals Index GPT-6 Astra #5 · 65.02%
See charts 3 benchmarksUpdated 19 Oct 2026LegalDrafting, extraction, legal reasoning and US legal research.
  • Legal Benchmarks leaderboard GPT-6 Astra #3 · 87.6%
  • LegalBench GPT-5.6 Sol #10 · 86.97%
  • Legal Research Bench GPT-5.6 Sol #11 · 48.08%
See charts 3 benchmarksUpdated 21 Oct 2026Health & medicalPhysician-graded health conversations, medical coding and exam questions.
  • HealthBench (independent run) GPT-6.1 Sol (max) #3 · 58.5%
  • MedCode GPT 5.1 #13 · 52.73%
  • MedQA o1 #1 · 96.52%
See charts 2 benchmarksUpdated 19 Oct 2026FinanceAnalyst research on SEC filings and long credit agreements.
  • Finance Agent GPT-6.1 Sol #13 · 61.82%
  • CorpFin v2 GPT 5.5 #9 · 68.42%
See charts 2 benchmarksUpdated 1 Sep 2026CodingFixing real GitHub issues and solving fresh programming problems.
  • SWE-bench Verified GPT-5.6 Sol #3 · 96.2%
  • LiveCodeBench GPT 5.2 Codex #9 · 87.99%
See charts 2 benchmarksUpdated 19 Oct 2026Agents & tool useMulti-step tasks an agent completes alone in a terminal.
  • Terminal-Bench 4.0 GPT-6 Astra #3 · 59.60%
  • Terminal-Bench 2.1 GPT-6 Astra #2 · 87.27%
See charts 4 benchmarksUpdated 21 Oct 2026Reasoning & knowledgeExpert exams, graduate science, contest maths and broad knowledge.
  • Humanity's Last Exam GPT 6 Astra #1 · 54.80%
  • GPQA Diamond GPT-5.6 Sol #2 · 95.20%
  • AIME GPT 5.2 #2 · 96.88%
  • MMLU Pro GPT-5.6 Sol #16 · 89.10%
See charts 1 benchmarkUpdated 20 Oct 2026Chat preferenceWhich answers real people prefer in blind side-by-side votes.
  • LMArena Text Arena GPT-5.6 Sol (xHigh) #21 · 1484
See charts 1 benchmarkUpdated 9 Jan 2026MultilingualReasoning in languages other than English.
  • MGSM GPT 5.2 #6 · 94.00%
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.