Expired50% confidenceFactExact time
A prompt tuned for GPT-4 may fail on Claude, and a prompt tuned today may fail after a model update.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026(expired)
Sources
Harness Engineering in Practice: Taming AI Agents with Hermes Agent
bilibili赋范课堂6/18/2026
Related Claims
Partially VerifiedGPT-5.5 and Claude 4.7 mentioned in the article do not currently exist; these are fictitious model version numbers.75% similarUnverifiedGPT-4 in function calling mode can comply with schema constraints at nearly 100%, while less capable models may produce malformed output or missing fields.74% similarVerifiedClaude 3.5 Sonnet is on par with GPT-4o across multiple benchmark tests.74% similarUnverified针对GPT-4优化的Prompt未必适用于Claude，源于各模型在预训练数据分布、RLHF偏好及安全过滤策略上的根本差异73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/44464API
curl https://kongchang.com/api/v1/knowledge/claims/44464MCP
get_claim(id=44464)