Unverified50% confidenceBenchmarkExact time
With 8 users each at 128K context, cached subsequent turns need only 0.2 to 2.5 seconds for the first token
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/9/2026
First Seen
Valid until: 2/7/2027
Sources
Related Entities
Related Claims
UnverifiedIn the long-context archiving experiment, 8 requests each input approximately 261,000+ tokens and each completed approximately 1,022 outputs; the total throughput of the joint decoding window was 214.85 tokens per second, approximately 26.86 per request, but the window was only approximately 2.28 seconds and prefilling was serial or interleaved.78% similarUnverifiedMicro-drama episodes typically run 60 to 90 seconds and end on a strong cliffhanger to drive users to keep unlocking.75% similarUnverifiedThe worst case is 8 users simultaneously sending brand-new 128K prompts, with everyone waiting about 2.5 minutes to get the first token74% similarUnverifiedUnder the same experimental setup, the total time to generate 1016 tokens drops from 14.66 seconds to 5.03 seconds.74% similarUnverifiedLong context has little effect on generation speed: after 4K context the median time per output token is 7.6 ms, and after 256K it is 8.3 ms; but the first read of a 128K-token prompt takes 18.33 seconds, so prefix caching is critical.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1150589API
curl https://kongchang.com/api/v1/knowledge/claims/1150589MCP
get_claim(id=1150589)