evals.report
BenchmarksLabsCompareRun guidesIn the wild
AnthropicClaude Haiku

Claude Haiku 5.5

Anthropic · Claude Haiku. Released Oct 7, 2026.

Claude Haiku 5.5 is a model from Anthropic in the Claude Haiku family, released Oct 7, 2026. evals.report tracks 8 reported Claude Haiku 5.5 benchmark scores across ARC-AGI-2, ARC-AGI-1, FrontierCode, FrontierCode 2.0, AA-Omniscience v1.1: Knowledge and Hallucination Benchmark, AA-Omniscience: Knowledge and Hallucination Benchmark, Humanity's Last Exam, SWE-bench Pro — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.

8 results

Benchmark results 8

Compare this model
BenchmarkCategoryScoreMetricStatusDate
ARC-AGI-2Reasoning61.25%accuracyOfficialOct 7, 2026Details
ARC-AGI-1Reasoning87.5%accuracyOfficialOct 7, 2026Details
FrontierCodeCoding46.4%weighted score (Main)OfficialOct 7, 2026Details
FrontierCode 2.0Coding28.9%weighted score (Main)OfficialOct 7, 2026Details
AA-Omniscience v1.1: Knowledge and Hallucination BenchmarkReasoning12AA-Omniscience Net ScoreOfficialOct 7, 2026Details
AA-Omniscience: Knowledge and Hallucination BenchmarkReasoning11AA-Omniscience IndexOfficialOct 7, 2026Details
Humanity's Last ExamReasoning45.9%accuracyVerifiedOct 7, 2026Details
SWE-bench ProCoding64.8%% resolvedUnverifiedOct 7, 2026Details