Cursor Max Mode: Legacy Plans, Cost, and Context

Max Mode lets a supported model read beyond its default context limit. Context means all the material sent with a request: the prompt, selected code, retrieved snippets, rules, tool results and conversation history. A larger window helps when a task crosses many files, or when a long session would otherwise lose something important.

The toggle changes how much the model can read, and nothing about how well it reasons. A weak plan, irrelevant files or a missing source still produce a bad result inside a large window. Start with good retrieval and an explicit scope, and add context only when the task needs it.

Only if it is a legacy request-based plan, the pricing Cursor used before it moved to usage-based pricing. Those plans counted usage in requests rather than tokens. Plans that meter tokens don't have Max Mode, whether they are current or legacy.

Check how your account is billed before you look in the model picker. A forum screenshot from someone else's plan is not reliable setup guidance for yours.

Larger context is still available on plans that meter tokens, where the model and your plan offer it. You choose it through the model's context option, and Cursor's model table mentions Max Mode only in notes about legacy request-based plans.

For team documentation, describe both paths in one sentence: legacy request-plan users may see Max Mode, and everyone else chooses context with the model. That keeps an internal guide useful during a migration, and support won't treat two valid interfaces as a bug. Record the plan beside any cost example, because the same model name doesn't prove the same billing path.

First confirm that the account is on a legacy request-based plan and that the chosen model supports Max Mode. The model picker is the source of truth for the session.

1. 1Open the model selector in the chat or Agent panel you are using.
2. 2Choose a supported model and check the context options shown for your plan.
3. 3Toggle Max Mode on for the task that needs the larger window. Models that require Max Mode turn it on automatically when you select them.
4. 4Turn it off when you no longer need it, because the setting stays on across conversations.
5. 5Review the usage after the task before you make it a team habit.

On any other plan, use the same reasoning but choose the context option on the model. Don't search for a hidden Max switch, and don't change plans only to reproduce an older interface. The question that matters is how much context the task needs.

Use extended context when the information a task needs is spread out and can't be retrieved reliably in smaller pieces. Examples are a cross-layer migration, a refactor with callers across packages, or an investigation whose evidence sits in code, tests, logs and configuration. A long file alone isn't enough reason.

To test it, rerun one representative task with a clean prompt and the smallest context that should work. If the narrow run misses a known dependency and the wider run finds it without extra correction, the larger window solved a real problem. If both pass, keep the smaller choice. If both fail, look at retrieval, instructions or model fit before you add context again.

On a legacy request-based plan, Cursor bills Max Mode at the selected model's API rate plus 20%.

The total still depends on the input, cache and output tokens a request uses, so no single price covers a Max chat. A long agent run can contain many model calls, each with a different amount of context.

The 20% surcharge is only part of the difference. A larger window can send far more input on every turn, so token volume can matter more than the multiplier. Check the usage dashboard after a representative task, and base team guidance on measured runs instead of a fixed request conversion.

For a team baseline, record the model, context choice, reasoning effortA model dial (high / medium / low) for how hard the model thinks before responding, set separately from which model and how fast it runs. Press Enter for the full definition., Fast state, number of agent turns, useful outcome and final usage. Compare at least a few tasks from the real repository. A policy like 'use Max for large refactors' is too broad to apply. A measurable trigger is easier to teach and review, such as 'standard context keeps missing a required package after explicit selection'.

These controls solve different problems. Context changes how much material a model can receive, and reasoning effortA model dial (high / medium / low) for how hard the model thinks before responding, set separately from which model and how fast it runs. Press Enter for the full definition. changes how much it deliberates. Fast uses priority processing for the same model, at a higher rate where offered. Cursor Router picks a model for each request on Teams and Enterprise by Cost, Balance or Intelligence, and Router Intelligence is not a context-window setting.

Set each one on its own merits. A hard algorithm may need high reasoning with modest context. A broad mechanical migration may need more context with ordinary reasoning, and an interactive debugging session may justify Fast.

Keeping them separate also speeds up troubleshooting. If the answer ignores a file, check context and retrieval. If it sees the facts but can't solve the logic, change the reasoning effortA model dial (high / medium / low) for how hard the model thinks before responding, set separately from which model and how fast it runs. Press Enter for the full definition. or the model. If a good result arrives too slowly, test Fast, and if a team wants automatic model choice, tune Router.

Change one control per run. Turning every control up at once makes the bill larger and doesn't show which change helped.