Unverified50% confidenceTradeoffExact time
The boundaries of RAG include: retrieval hit rate depends on chunking granularity and embedding model quality, and too many entries will squeeze the context budget, potentially leading to 'remembering the characters but not the plot'.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRAG依赖精确的向量检索推理成本低但受限于检索质量，超长上下文窗口避免检索失配但KV缓存显存占用随上下文长度线性增长导致推理延迟和成本上升73% similarUnverifiedStuffing massive volumes of documents in their entirety into an oversized context window faces problems of cost, latency, and attention degradation (needle in a haystack), and the results may not be better than precise retrieval.72% similarUnverifiedRetaining reasoning context may incur additional token usage, and monitoring should be in place in cost-sensitive scenarios.70% similarUnverifiedThe most common problem in on-device inference is memory, as context length silently inflates the KV cache until it collides with the weights.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1030302API
curl https://kongchang.com/api/v1/knowledge/claims/1030302MCP
get_claim(id=1030302)