Cursor Auto Cost at $1.25/$6: Router Pricing and the $0.25 Token Surcharge

Cursor now describes three Auto modes. Auto Cost uses fixed rates of $1.25 per million input tokens, $0.25 for cache reads, and $6 per million output tokens. Auto Balance and Intelligence use the routed model API rate, and Teams or Enterprise third-party requests can add a $0.25-per-million Cursor Token Rate.

The factual baseline comes from Cursor's current models and pricing documentation. Product behavior and prices can change, so confirm the current page before making a purchase or locking an annual budget.

The budgeting mistake is to look at one visible rate and treat it as the whole system. Coding work crosses planning, context retrieval, tool calls, file edits, execution, validation, repair, and review. Every stage can add tokens, runtime, or people. A useful model preserves those layers and then rolls them into cost per accepted result.

What Matters for the Cost Model

• Auto Cost is priced independently of which underlying model the router selects.
• Balance and Intelligence follow the chosen model published API rate rather than one fixed blended rate.
• The Cursor Token Rate applies to third-party selections and qualifying routed requests on business plans.
• First-party Cursor models and Auto Cost are documented as exempt from that additional token rate.

These facts change the shape of spend, but they do not determine whether the product is economical. A higher-priced run can be the cheaper outcome if it merges on the first attempt and needs little review. A low-cost run becomes expensive when it wakes repeatedly, occupies compute while idle, or produces a patch that experts must rewrite.

Build the Equation Before the Forecast

For Auto Cost, a request with 600,000 input tokens, 200,000 cached input tokens, and 80,000 output tokens costs $0.75 plus $0.05 plus $0.48, or $1.28. For a routed third-party model, calculate its input and output charge first, then add the applicable token-rate line according to the plan rules.

Keep cached input, uncached input, and output separate because their rates can differ by an order of magnitude. Add tool fees, search calls, image or screenshot processing, and regional premiums only when the workflow actually uses them. For runtime, model billing minimums and rounding explicitly. A spreadsheet that assumes perfect per-second billing can understate a block-priced service.

A Worked Monthly Example

A team processing 120 million third-party tokens a month would see a $30 surcharge at $0.25 per million, before model charges. That is small beside premium output rates but material beside an ultra-cheap model. If the underlying model bill is $60, the surcharge raises effective spend by 50%; if it is $1,200, it adds only 2.5%.

Run a sensitivity range around the estimate. Use a low, expected, and high case for acceptance rate, output length, retry count, and runtime. Those variables usually move the answer more than a tiny difference in input-token price. The high case should represent a difficult but plausible sprint, not an impossible catastrophe.

Controls That Preserve the Savings

• Log the routed model for every request. Make this visible in the run record so the team can connect the control to dollars and accepted output.
• Separate model rate and platform surcharge. Make this visible in the run record so the team can connect the control to dollars and accepted output.
• Benchmark Cost mode on real repository tasks. Make this visible in the run record so the team can connect the control to dollars and accepted output.
• Set allowlists so routing cannot jump to an unsuitable tier. Make this visible in the run record so the team can connect the control to dollars and accepted output.

Give every automated run a durable identifier that follows retries, branches, builds, and review. Without that join key, invoices remain disconnected from engineering outcomes. Aggregate by task class, repository, team, model, and trigger so a change in workload mix does not masquerade as a price change.

The Failure Mode to Watch

A router can improve average economics while making individual requests less predictable. Averages also hide mix shifts: one week of difficult migrations can route more traffic to frontier models. Forecast by task class and monitor cost per accepted change, not only the blended cost per token.

Set a stop condition before the run begins. Repeated failure on the same test, no material repository change, a spend ceiling, or a deadline can all trigger escalation. Stopping is not failure; it protects the budget and gives a human the evidence needed to change the approach instead of financing another identical loop.

How to Decide

Auto Cost favors predictability and high-volume everyday work. Balance or Intelligence can be rational when a stronger model materially increases first-pass success. The correct comparison includes retries and review time; a low fixed token rate that produces more failed attempts is not automatically cheaper.

Start with a bounded pilot and preserve a control group using the current process. Track model dollars, runtime dollars, tool charges, reviewer minutes, acceptance, cycle time, and post-merge defects. After enough comparable tasks, calculate total cost per accepted result. Use the AI Cost Calculator for the token component, then add the operational layers the API invoice does not show.