Unverified50% confidenceOpinionExact time
In interactive dialogue and writing scenarios, decode speeds above 20 tokens/s are generally smooth, while speeds below 10 tokens/s produce a noticeable sense of waiting.
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA query that takes 2 seconds in standard mode may become 25 seconds in deep reasoning mode, while consuming 5 to 10 times the tokens.69% similarUnverifiedFor a chatbot, 40 seconds is slow, but for a letter one has been procrastinating on because one does not want to read it, it is fast enough; the right question is not "is it fast" but "is it fast enough for this task".68% similarUnverifiedAccording to the demo description, the Decisions API achieves server-side response times below 100 milliseconds in text classification scenarios68% 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.67% similarUnverifiedA generation speed of 0.5 tokens/second is comparable to the speed of human thinking output.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1062230API
curl https://kongchang.com/api/v1/knowledge/claims/1062230MCP
get_claim(id=1062230)