Atlas

Models

← All models

Claude Opus 5

Anthropic · 2026-07-24

Claude Opus 5 is Anthropic's July 2026 flagship Opus model for long-running agents, coding, and professional work, with adaptive thinking and a 1M-token context window.

Benchmark scores

BenchmarkScore
A-Fantasia (Valid Attempts with Retries)29.7
A-Fantasia Chess (Valid Attempts with Retries)27.0
A-Fantasia Cube (Valid Attempts with Retries)47.0
A-Fantasia Spelling (Valid Attempts with Retries)15.0
AA-AnalystAgent (pass@1)63.8
AA-AnalystAgent (pass@5)73.8
AA-AnalystAgent (pass^5)53.8
AA-Briefcase Elo1720
AA-Briefcase v1.1 Elo1655
AA-LCR70.0
AA-LCR v1.182.0
AA-Omniscience Index37.1
ActiveVision10.6
ActiveVision with tools68.2
AdvancedIF80.5
AdvancedIF (rubric mean score)95.5
Agent Red Teaming k=100 (CAIS)43.3
Agents' Last Exam (ALE)32.2
Agents' Last Exam ALE-CLI Pass Rate29.5
Agents' Last Exam ALE-CLI Score57.3
Agents' Last Exam Full-Spectrum Pass Rate29.1
Agents' Last Exam Full-Spectrum Score47.9
Agents' Last Exam Last-Exam Pass Rate13.2
Agents' Last Exam Last-Exam Score25.8
Agents' Last Exam Near-Term Pass Rate49.3
Agents' Last Exam Near-Term Score80.1
Agents' Last Exam Overall Score55.9
Agents' Last Exam Unlicensed Full-Spectrum Pass Rate30.0
Agents' Last Exam Unlicensed Full-Spectrum Score48.8
Agents' Last Exam Unlicensed Last-Exam Pass Rate16.7
Loading Atlas data…