Cursor's current documentation describes separate monthly usage pools for Cursor models and third-party models. Which pool a request uses depends on the selected model and your plan, and model choice affects how quickly included usage is consumed.
Check which usage pool is being consumed.
Cursor says most plans include a Cursor Models pool and an Other Models pool; plan entitlements differ. Its dashboard shows real-time usage, remaining allowance, and any on-demand charges. Usage resets monthly with the billing cycle, and unused usage does not roll over.
Max Mode is a legacy-plan feature.
Current usage-based plans do not include Max Mode. Cursor documents it for legacy request-based plans, where it extends the model's context window and is billed at the model's API rate plus 20%. Do not assume this behavior applies to every current plan.
What to do when usage is running low.
- Stop adding unrelated files and requests to the current thread.
- Preserve the changed files, diff, errors, decisions, and verification state.
- Separate confirmed facts from claims the next agent must re-check.
- Restart in Cursor or switch tools with a forward brief instead of pasting the whole transcript.
The portable unit is the work.
Cursor is useful while the session is alive, but the project should not depend on one editor's context window or billing meter. ShardStitch builds the continuation packet from local project state so the next step can happen in a fresh Cursor chat or another AI coding tool.
Cursor usage limits FAQ
Does Max Mode apply to current plans? No. Cursor says current usage-based plans do not include Max Mode. It remains available on legacy request-based plans and is billed at the model API rate plus 20%.
How do I move a Cursor task to another AI tool? Capture the goal, changed files, diff, failed attempts, verification state, and next action, then start the next tool from that continuation packet.