Unverified50% confidenceOpinionExact time
A higher physical ceiling on the output window does not mean the effective output length increases proportionally; a model that can generate 1 million tokens but starts hallucinating at 200K tokens has less practical value than one that stably outputs 300K tokens.
1
Sources
50%
Confidence
Long-term
Relevance
10/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedEven well-prompted, schema-constrained models with 8 billion to 14 billion parameters still have an error rate of a few percent on real-world inputs—this is a rate to design around, not a bug to eliminate.68% similarUnverified对万亿参数模型而言，处理长上下文时KV Cache的显存开销甚至可能超过模型权重本身66% similarVerifiedThe emergent property of CoT manifests only when model parameter count exceeds approximately 100 billion, at which point it significantly improves performance, while being almost ineffective on smaller models.66% similarUnverifiedThe single-response output limit for most large language models is concentrated in the range of several thousand to tens of thousands of tokens.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/965084API
curl https://kongchang.com/api/v1/knowledge/claims/965084MCP
get_claim(id=965084)