• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
  • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Users debate the impact of Kimi’s alleged routing to Claude

    One user dismisses the scale of the allegation, citing 300,000 requests over 10 days. Another argues that which queries were routed could matter more than the volume.

    SR
    DJ
    YA
    3 Sources, ,

    TLDR

    One user mocks claims that Claude was behind Kimi’s performance, citing an article’s figure of 300,000 requests over 10 days. Another argues that routing the toughest 1% of queries to Claude could have a meaningful impact, claiming the models perform similarly on the other 99%, characterized as day-to-day requests. That 1% is a hypothetical routing strategy, not a reported breakdown.

    Combined views

    37.8K

    3 Sources, first seen 19d ago

    Combined views

    37.8K

    3 Sources, first seen 19d ago

    478 likes
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    19d ago
    first seen 19d ago
    478 likes
    16 comments
    41 saves
    94 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    16 comments
    41 saves
    94 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    3 Sources

    @zehavocRT @xz_keg: It is actually impossible because Kimi and Deepseek offer reasoning traces to users, the user can detect instantly if it routes…
    @yaghsizian> “wowa kimi was fooling everyone and faking it’s performance by routing to claude! It was claude all along!!!” > checks article > 300,000 requests over 10 days lol wait till you find out I processed a billion tokens the other day people really have 0 concept of scale
    @cloneofsimoIf Kimi bros iq was non-zero, it would have the claude routed for toughest 1% of the queries, as for 99% of normies queries (such as day-to-day stuff) kimi and claude does similarly. Then such intervention is nontrivially impactful

    3 Sources

    @zehavocRT @xz_keg: It is actually impossible because Kimi and Deepseek offer reasoning traces to users, the user can detect instantly if it routes…
    @yaghsizian> “wowa kimi was fooling everyone and faking it’s performance by routing to claude! It was claude all along!!!” > checks article > 300,000 requests over 10 days lol wait till you find out I processed a billion tokens the other day people really have 0 concept of scale
    @cloneofsimoIf Kimi bros iq was non-zero, it would have the claude routed for toughest 1% of the queries, as for 99% of normies queries (such as day-to-day stuff) kimi and claude does similarly. Then such intervention is nontrivially impactful