Unverified50% confidenceFactExact time
The keep_alive=10m parameter tells Ollama to retain model weights in memory for ten minutes after a request ends, avoiding the latency of reloading the model on subsequent calls.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/11/2026
First Seen
Valid until: 1/9/2027
Sources
Related Entities
Related Claims
UnverifiedAn identical request to the same model immediately following took only approximately 1.6 seconds, nearly 10 times less than the first request.64% similarUnverifiedDurable Objects automatically hibernate when there are no requests for a few seconds, saving memory state via checkpointing, dropping CPU usage to zero, with wake-up time ranging from 10–100 ms depending on state size.63% similarUnverifiedThe author considers a model usable only if it achieves at least 10 tokens per second.60% similarUnverifiedWith all three optimizations stacked, generation speed stabilizes at 9 to 10 tokens per second, but time-to-first-token for a 10,000-character long document still requires approximately 329 seconds.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/999875API
curl https://kongchang.com/api/v1/knowledge/claims/999875MCP
get_claim(id=999875)