Cursor announced changes to Auto pricing and usage limits effective August 24, 2026. Pricing will shift from a flat rate to model-based pricing, with most requests costing more than before, but included usage limits will increase. Users need no action, and Auto will continue to route requests to the best model.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
Starting August 24, 2026, Cursor's Auto mode will switch from a flat rate to model-based pricing, increasing costs for most requests while also increasing included usage limits. This change has sparked concerns in the developer community about cost transparency and debugging experience.
According to Cursor's official announcement (shared via X), starting August 24, 2026, Auto mode pricing will switch from a flat rate to model-based pricing. This means each request's cost will be calculated based on the model used, rather than a fixed uniform price.
The announcement explicitly states that most requests will cost more than under the old flat rate, because Cursor needs to route everyday tasks to the cheapest suitable model to maintain cost efficiency. This implies that while overall costs may rise, Cursor still attempts to control expenses through intelligent routing.
The announcement also mentions that included usage limits will increase and already apply to the current billing cycle, requiring no action from users. This means existing users will immediately benefit from higher limits, while the new pricing structure takes effect on August 24.
Notably, the announcement emphasizes that Auto mode's functionality itself remains unchanged: it still routes each request to the most suitable model and maintains the best value per dollar of intelligence. This indicates Cursor aims to sustain its core value proposition while adjusting pricing.
In discussions on X, users have mixed reactions to this adjustment. One user asked whether the free plan still includes Auto's free allowance, but the announcement does not clarify specific changes to the free plan, so there is currently no official answer.
Another user pointed out that model-based pricing means the 'cheap path' is no longer cheap by default, which will be a directly noticeable change. Others worry that routing everyday tasks to the cheapest model might 'silently degrade the quality of debugging sessions' and hope Cursor will disclose which model is actually used for each request to increase transparency.
Currently, the announcement does not provide a specific price list or pricing details for each model, nor does it state whether the free plan is affected. This information is crucial for users to assess cost impact but has not been released.
Additionally, the statement that 'most requests will cost more' is based on Cursor's official wording, but the exact increase and applicable scenarios remain unclear. User demands for model transparency have also not received an official response.
This report is based on content shared on X, with the original source being Cursor's official announcement. This information is a source claim and has not been independently verified. Specific price details and free plan policies have not been disclosed, so related inferences should be treated with caution.
Cursor Auto's pricing adjustment will increase costs for most requests but also increase included usage limits. Users should monitor the official detailed price list and assess whether their usage patterns are affected.
Primary report
Primary source