Verified70% confidenceFactTime unknown
Even top-tier models like GPT-4 can have hallucination rates of 15%–30% in specific domains.
4
Sources
70%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA 2024 study from UC Berkeley found that in complex multi-step tasks, GPT-4-level models exhibited tool-call hallucinations at a rate as high as 12%–18%.80% similarUnverifiedEven GPT-4-level models cannot fundamentally eliminate hallucinations.79% similarUnverifiedMainstream large language models including GPT-4, Gemini, and Grok have all had hallucination behavior documented by researchers75% similarUnverifiedThe capability boundary of the local 27B model is roughly equivalent to the level between early GPT-3.5 and GPT-4.74% similarUnverified2023年多项研究表明，GPT-4在回答医学问题时虚构参考文献的比例可高达30%-70%74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/6112API
curl https://kongchang.com/api/v1/knowledge/claims/6112MCP
get_claim(id=6112)