Frontier Log

Claude Haiku 5.5

Claude Haiku 5.5 on Humanity's Last Exam (no tools) by release date: 45.9%, up from Haiku 4.5's 10.2%. It is above Opus 4.6 (40.0%) from eight months earlier and Sonnet 5 (43.2%), and below Opus 4.8 (49.8%), Sonnet 5.5 (56.9%) and Opus 5.5 (64.4%)

Download

Humanity's Last Exam, no tools

ModelScoreSources
Claude Sonnet 4.517.7%AnnouncementFigure 2.16.A 'Humanity's Last Exam (HLE)', panel 'Without Search', bar 'Sonnet 4.5' (page 32)
Claude Haiku 4.510.2%AnnouncementPerformance table, row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Haiku 4.5', 'no tools'
Claude Opus 4.530.8%AnnouncementFigure 2.16.A 'Humanity's Last Exam (HLE)', panel 'Without Search', bar 'Opus 4.5' (page 32)
Claude Opus 4.640.0%Announcementbenchmark table (image), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 4.6', 'without tools'
Claude Opus 4.849.8%Announcementbenchmark table (image), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 4.8', 'no tools'
Claude Sonnet 543.2%Announcementbenchmark table (image), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Sonnet 5', 'no tools'
Claude Opus 5.564.4%AnnouncementSection 8.1 evaluation summary table, row 'Humanity's Last Exam, No tools', column 'Claude Opus 5.5'
Claude Sonnet 5.556.9%AnnouncementSection 8.1 evaluation summary table, row 'Humanity's Last Exam, No tools', column 'Claude Sonnet 5.5'
Claude Haiku 5.545.9%AnnouncementPerformance table, row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Haiku 5.5', 'no tools'

Claude Haiku 5.5 on the timeline