Unverified50% confidenceBenchmarkExact time
The perplexity of the GPT series of models on Wikipedia text has decreased from approximately 29 for GPT-2 to an even lower level for GPT-4.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/6/2026
First Seen
Valid until: 11/4/2026
Sources
马尔可夫链熵率详解：公式推导与序列建模应用
Related Claims
UnverifiedThe capability improvement from GPT-4 to GPT-4o is notably smaller than the leap from GPT-3 to GPT-4, indicating diminishing returns from the Scaling Law79% similarVerifiedGPT-4经历了从GPT-4到GPT-4o再到GPT-4.1的版本迭代78% similarUnverifiedThe context limit in the GPT-4 era was 8,192 tokens.77% similarPartially VerifiedAll three GPT-5.6 models achieved a perfect score of 10 on the long-horizon agent fine-tuning task, while GPT-5.5 and Opus 4.7 could only score two or three points on that task.76% similarUnverifiedCompared to previous generations, GPT-5.6 has significantly fewer bugs and reviews the overall structure rather than just local code.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/697621API
curl https://kongchang.com/api/v1/knowledge/claims/697621MCP
get_claim(id=697621)