Claude 4.5 Haiku benchmark ranking on every AI leaderboard

Name: Claude 4.5 Haiku
Author: Anthropic

closed weights · #40 overall, podium score 3.3 · top 10 on 1 of 12 · released Oct 1, 2025

12 of 12 boards

Rank	Board	Category	Score	Of
6	BFCLAgents and tools · as “Claude-Haiku-4-5-20251001 (FC)”	Agents and tools	68.7%	81
11	SWE-Bench ProCoding · as “Haiku 4.5 (Claude Code) xhigh”	Coding	25.5%	11
23	MultiChallengeChat · as “claude-haiku-4-5-20251001-thinking”	Chat	50.5%	29
32	Document ArenaVision · as “claude-haiku-4-5-20251001”	Vision	1,420	38
34	MCP AtlasAgents and tools · as “claude-haiku-4-5”	Agents and tools	40.2%	34
55	ARC-AGI-2Reasoning · as “Claude Haiku 4.5 (Thinking 32K)”	Reasoning	4.0%	78
69	AA IntelligenceReasoning · as “Claude 4.5 Haiku (Reasoning)”	Reasoning	16.9	100
88	WebDev ArenaCoding · as “claude-haiku-4-5-20251001”	Coding	1,329	100
91	GPQA DiamondReasoning · as “claude-haiku-4-5-20251001_32K”	Reasoning	71.2%	100
93	Epoch ECIReasoning · as “Claude Haiku 4.5”	Reasoning	142.4	100
100	Text ArenaChat · as “claude-haiku-4-5-20251001”	Chat	1,415	100
13	SWE-bench VerifiedCoding · stale since 2026-02-26 · as “Claude 4.5 Haiku (high)”	Coding	66.6%	39

Category ranks and more models

Scores as published by each board on the capture date. Model names and logos belong to their owners; logos via logo.dev.