Max Mode is documented as available only on legacy request-based plans (API rate + 20% surcharge); current plans use selectable context windows instead.
Changelog
Documentation updates
Models & pricing
Max Mode is documented as available only on legacy request-based plans (API rate + 20% surcharge); current plans use selectable context windows instead.
Model pages (Claude Sonnet/Opus and related) describe large context windows without Max Mode, with base per-token rates applying above default context.
Cloud Agents
Cloud Agents no longer always run in Max Mode; you can select context window size for supported models (larger windows can increase token usage and cost).
Automations use each model’s maximum supported context window (no context-window toggle).
Mobile: pick a worker per run—Cloud machine, Self-Hosted Pool, or My Machines; runs use the model’s max context window.
API: model.params documents context window size (alongside reasoning effort) instead of max mode.
Default model setting: any model available for cloud agents (no Max Mode requirement).
CLI
/max-mode is documented for legacy request-based plans only; the Max Mode section was removed from the CLI overview.
