Unverified80% confidenceFactTime unknown
根据Anthropic官方数据，Haiku 4.5在SWEbench上击败了Sonnet 4，甚至超过了GPT-5
1
Sources
80%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026
Sources
Claude Haiku 4.5实测：速度虽快，性价比不敌GPT-5 Mini
bilibili一摩尔炸鸡翅
Related Entities
Related Claims
UnverifiedAnthropic声称Sonnet 4.5在编码任务上超越了GPT-573% similarUnverifiedGLM-4.7在浏览器能力测试中超过DeepSeek V3.2和GPT-5.1，明显领先Claude Sonnet 4.571% similarUnverified文章声称Claude Haiku 4.5在多项核心基准测试中击败了GPT-5和Gemini 2.5 Pro70% similarUnverified在SaaS落地页生成测试中，四个模型（Claude Haiku 4.5、Claude Sonnet 4.0、Gemini 2.5 Pro、GPT-5）全部成功完成任务，但Haiku 4.5生成速度最快70% similarUnverifiedGPT-5.6的三档模型布局对标Anthropic的旗舰Opus、中型Sonnet和高效Haiku66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/24313API
curl https://kongchang.com/api/v1/knowledge/claims/24313MCP
get_claim(id=24313)