Unverified65% confidenceFactTime unknown
GPT-5 Mini在同类测试中表现远优于Claude 4.5 Haiku
1
Sources
65%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026
Sources
Claude 4.5 Haiku实测翻车：编码能力全面溃败，性价比被竞品碾压
bilibili烟神殿殿主
Related Entities
Related Claims
UnverifiedClaude 4.5 Haiku在高中级别编程任务上几乎能和GPT-5打平手83% similarUnverifiedClaude Haiku 4.5相比GPT-4o mini等同级别模型在定价上仍有明显优势82% 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 Pro81% similarPartially Verified本次测评对Claude Opus 4.8、GPT 5.5、MiniMax M3、Mimo 2.5 Pro和DeepSeek V4 Pro五个模型进行了横向比较79% similarUnverifiedGPT-5.6 SOUL版本多项测试表现优于对标模型Fable 5，定价更具竞争力79% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/24357API
curl https://kongchang.com/api/v1/knowledge/claims/24357MCP
get_claim(id=24357)