February 5th, 2026

Updates for February 5, 2026

  • Added strength training support — select Strength alongside Cycling, Running, and Swimming in Settings → Training to include gym sessions in your weekly plan. Existing strength/gym sessions in your Intervals.icu calendar no longer block endurance workout generation.
  • Added Czech and Slovenian language support — switch languages in Settings → Account
  • Calendar improvements — navigate to past months, larger icons/text for better readability, TSS heat-map backgrounds showing training load intensity, and inline legend to save space
  • Fixed swimming workouts showing pool length as duration — "4x 25m" was being parsed as "4x 25 minutes" instead of "4x 100m" because Intervals.icu interprets "m" as minutes
  • Fixed threshold workouts generating at 75% FTP instead of 95-105% — added explicit power zone guidance for all cycling, running, and swimming workout types
  • Improved settings tab selector visibility on mobile — added clearer border, shadow, and dropdown indicator
  • Fixed aggressive weekly TSS targets for goal-based training — now caps CTL ramp at 6 points/week instead of trying to reach target too quickly
  • Fixed calendar not updating after manually uploading activities to Intervals.icu — calendar cache now refreshes when activities are processed
  • Fixed race pace plan failing to load for some events — improved handling of special characters in AI-generated race strategies
  • Improved race strategy in calendar event details — now uses same full context as race day email (fitness, wellness, phase) for better advice
  • Fixed AI advice treating stale wellness data as current — prompts now indicate when data is from yesterday or older
  • Fixed post-workout analysis for FTP tests — the AI now recognizes test workouts (FTP test, ramp test, time trial) and analyzes Normalized Power and Intensity Factor instead of average power, which was incorrectly lowered by warmup and cooldown
  • Fixed Goal Setup wizard showing stale CTL from months ago — wizard was reading oldest fitness data instead of most recent, causing incorrect AI recommendations and slider ranges