CursorAI/Q/Why does Cursor have token limits that are lower than the model's advertised context limits?

• Why does CursorAI have token limits that are lower than the model's advertised context limits?

  • AI Answer

    • Answer/Official from Max Mode on legacy plans and Models & Pricing: Cursor publishes a Default context and a separate Max context per model. The default is intentionally lower than the provider's full window; Max Mode (legacy request-based plans) or an explicit context-size choice in the model picker (usage-based plans) opts into the larger window.
    • Short answer: Cursor caps the default window below the model's advertised max for cost and product defaults — not because the model lacks capacity. Staff have also said Cursor may lower a model's usable limit based on how it performs at that size. The UI meter can look even smaller because system prompts, tools, reasoning, and output reserve share the same window.
    • Default vs max is a product choice. Cursor's models table lists both columns (e.g. many Claude models: default 200k, max 1M). Official Max Mode help: the default window "works well for most coding tasks"; use the larger window when a task needs more. Larger context burns more tokens (and some providers charge more above long-context thresholds).
    • Cursor may also cap below the provider max for performance. Forum staff (Condor): each model has a provider limit, and "We also may set a context limit to a model depending on its performance in non-Max mode." (Max Mode discussion)
    • How to get the larger window. On usage-based plans, pick the context size in the model picker (Max Mode as a separate toggle was removed for those plans; same sizes remain). On legacy request-based plans, toggle Max Mode — billed at API rate plus 20%.
    • Why the usage bar can still look short of even Cursor's limit. Reserved space for system/agent instructions, tool definitions, reasoning tokens, and the reply means the "available" user context is less than the configured window. That is separate from the default-vs-max policy above. (Forum thread on truncated available context)
    • Sources:
      • Max Mode on legacy plans (Cursor Docs)
      • Models & Pricing / docs overview models table — Default vs Max context columns
      • Does MAX mode provide additional benefits… (Cursor Forum — staff on default caps)
      • What does Cursor removing "Max Mode" mean… (context size selectable in model picker)