Expired50% confidenceBenchmarkExact time
On computer use and world knowledge tasks, Sonnet 5 has approached the level of Opus 4.8, but still lags notably behind Opus 4.8 in the domain of agentic programming
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/20/2026
First Seen
Valid until: 10/18/2026(expired)
Sources
Related Claims
UnverifiedClaude Sonnet 4.5 significantly outperforms the previous-generation Sonnet 4 model on software engineering benchmarks, and even surpasses the higher-positioned Opus model.77% similarUnverified测试者认为Opus 5是优秀的编程模型，能力与Sonnet 5差距不大且价格更低73% similarUnverifiedClaude Haiku 4.5 can already match the previous generation flagship model Sonnet 4 in programming performance69% similarUnverifiedIn SVG generation tasks, Sonnet 5's results for generating a BMW M4 CS were completely unsatisfactory, clearly trailing behind GLM, and the tester ranked Sonnet 5 below GLM 5.2.68% similarUnverifiedPerplexity's connector issue is especially pronounced when using the Sonnet 5 model.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/571452API
curl https://kongchang.com/api/v1/knowledge/claims/571452MCP
get_claim(id=571452)