Ready for Token Factory's November model changes. Nebius retires Nemotron 3 Super and Nano from Token Factory Serverless on November 9, during judging. Huddle now runs Super's work (Just ask's plans, weighing sources in fact checks, filling tables and charts, tidying memory, learning from what you say) on Nemotron 3.5 Super VL, Nebius's recommended replacement at the same price ($0.30 in / $0.90 out per million tokens).
Flows and assistants already saved with a retiring model keep working: Huddle asks the replacement in its place and prices it the same way.
Tested live on Token Factory today: Just ask planned a two-assistant team in 1.4 s (1.9 s on Super before), and a cart pick, memory learning and a table all came back correct in under 2 s. The live demo runs the new code.
