Unverified50% confidenceBenchmarkExact time
In an experiment with 6 sessions, 5 messages per session, and 600 KB per message (18.4 MB of input in total), 192.4 MB was written to disk
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027
Sources
Related Entities
Related Claims
UnverifiedStoring a 1-million-token conversation now requires less than 1 GB of memory, whereas larger previous-generation models required at least 3.7 GB.72% similarVerifiedFor a typical 7B model with 32 layers and 32 attention heads, storing the KV Cache for each token requires approximately 0.5 MB of VRAM, so generating 1,000 tokens requires an additional ~500 MB of VRAM71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1113685API
curl https://kongchang.com/api/v1/knowledge/claims/1113685MCP
get_claim(id=1113685)