Cursor's Auto Cost mode drops its flat per-token rate
Cursor's Auto Cost, Balance, and Intelligence modes now all bill at the routed model's own list price; the old flat $1.25 in / $6 out blended rate survives only as 'Legacy Enterprise Auto,' sunsetting September 7, 2026.
Auto Cost billed a fixed blended rate ($1.25 input/cache-write, $0.25 cache read, $6 output per 1M tokens) regardless of which model it routed to, for all individual, Teams, and Enterprise Auto usage. Auto Balance and Auto Intelligence billed at the routed model's own API rate.
All three Auto modes (Cost, Balance, Intelligence) bill at the selected/routed model's own list price. The flat blended rate is now labeled 'Legacy Enterprise Auto,' available only to Enterprise accounts still on it, and is scheduled to sunset on September 7, 2026.
Proof of change
Cursor (Anysphere)'s pricing pages, as we captured them on two dates.
Showing the whole page as captured — scroll either panel, or open it at full size.
Show 3 other pages we compared
Showing the whole page as captured — scroll either panel, or open it at full size.
Showing the whole page as captured — scroll either panel, or open it at full size.
Showing the whole page as captured — scroll either panel, or open it at full size.
- These prices come from our capture alone — they were not confirmed against an independent second source.
- 4 pages had no counterpart in the earlier capture and are not shown.
Both images are our own captures, taken on the dates shown. The pixels are unmodified — nothing is retouched, and where a region is highlighted the marker is drawn over the image, not into it. Open either capture to see it at full resolution.
Cursor’s Models & Pricing docs previously described Auto Cost as a fixed per-token rate that let users route requests without worrying about which underlying model got picked — a deliberate flat-rate safety valve against the variable credit-pool pricing. As of this capture, the docs now state plainly that “all Auto modes bill at the list price of the model each request is routed to,” collapsing Cost, Balance, and Intelligence onto the same billing mechanic (they still differ in which model each mode routes to, not in how the resulting usage is priced). The old flat rate remains available only to Enterprise accounts already on it, under the new label “Legacy Enterprise Auto,” and is set to sunset on September 7, 2026. Plan-level pricing was unchanged.
Cursor's Other Models pool cut Claude Sonnet 5 roughly 33% (input $3 to $2, output $15 to $10 per million tokens), GPT-5.6 Sol roughly 20-33% (input $5 to $4, output $30 to $20), and replaced Gemini 3.6 Flash with Gemini 3.7 Flash at about half the per-token rate (input $1.5 to $0.75, output $7.5 to $3.5) — all pass-through cuts. Cursor Models also gained Grok 4.6 and Grok 4.6 (Fast) alongside the existing Grok 4.5 line. Plan-level pricing (Pro $20, Pro Plus $60, Ultra $200, Teams $40/$120) was unchanged, and the annual-billing toggle was confirmed to apply a 20% discount (Pro $16/mo, Teams Standard $32/user/mo).