Cursor vs Claude Pro (Claude Code): limits compared
| Cursor | Claude Pro (Claude Code) | |
|---|---|---|
| Plan | Pro | Pro |
| Price $/mo | 20 | 20 |
| Metered in | Dollars of usage | Five-hour and weekly limits |
| Stated limits | At least $20/mo of third-party models at API prices plus a separate pool for Cursor's own models; overage continues at same rates | Claude Code included; at least 5x Free per five-hour session, with weekly limits shared with chat; extra use at API rates or bundles ($45 buys $50) |
| Resets | Monthly | Every 5 hours + weekly |
Cursor user reports
- “Cursor (auto) - 60M tokens / month ($20)” (User estimate from ccusage/Cursor usage stats) source
- “Composer with Opus burns through the $20 plan in days.” source
Claude Pro (Claude Code) user reports
- “Opus - 10M tokens / 5hr; Sonnet - 20M tokens / 5hr; Haiku - 40M tokens / 5hr” (User estimate (ccusage)) source
- “I can easily run out of tokens with 2 or 3 feature requests or debugging sessions.” source
Units differ between vendors and aren't directly comparable.