4 ms·

done pretty much inline with the price elo pareto frontier https://x.com/swyx/status/1912959140743586206/photo/1 https://x.com/swyx/status/1912959140743586206/p

• xnx 2y ago
  Love that chart! Am I imagining that I saw a version of that somewhere that even showed how the boundary has moved out over time?
  • swyx 2y ago
    https://x.com/swyx/status/1882933368444309723 https://x.com/swyx/status/1882933368444309723 https://x.com/swyx/status/1830866865884991999 https://x.com/swyx/status/1830866865884991999 (scroll up)
• oezi 2y ago
  So if I see it right flash 2.5 doesn't push the pareto front forward, right? It just sits between 2.5 pro and 2.0 flash. https://storage.googleapis.com/gweb-developer-goog-blog-assets/images/Gemini-2.5-Flash-price-to-performance-compariso.original.png https://storage.googleapis.com/gweb-developer-goog-blog-asse...
  • swyx 2y ago
    yeah but 1) its useful to have the point there on the curve if you need it, 2) intelligence is multidimensional, maybe in 2.5 flash you get qualitatively a better set of capabilities for your needs than 2.5 pro
  • jsnell 2y ago
    It does, that point in the tradeoff space was not available until now. Any model that's not dominated by at least one model on both axes will push forward the frontier. (The actual frontier isn't actually a straight line between the points on the frontier like visualized there. It's a step function.)