Expired85% confidenceFactTime unknown
Claude Haiku 4.5 has a lower false refusal rate than Sonnet 4.5 and Opus 4.1, making it one of the safest models Anthropic has released to date.
1
Sources
85%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedClaude Haiku 4.5 surpasses the larger Sonnet 4.5 on certain computer use task benchmarks.76% similarUnverifiedAccording to benchmark data released by Anthropic, Sonnet 4.5 is slightly superior to Claude Opus 4.1 in coding capability.75% similarUnverifiedAnthropic stated that Sonnet 4.6 performs on par with Opus 4.6 in terms of resistance to prompt injection.74% similarUnverifiedIn the weather website animation effect test, Claude Haiku 4.5's CSS animation detail performance surpassed that of Claude Sonnet 4.0.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/24451API
curl https://kongchang.com/api/v1/knowledge/claims/24451MCP
get_claim(id=24451)