Head to head / curated pair / cycle 04
Cursor vs Claude Code
Both products were given the identical 9 prompt vibeOps specification, 5 times each. Nothing on this page is an impression, all of it is a measurement from the same harness.
The strongest edit loop we measured. It keeps a large working set in view, so the tenancy constraint from prompt 3 survived every later change without being restated. As a coding agent it produces no interface of its own, which caps the design subscore: what you get is exactly the quality of the framework and components you point it at. Everything it writes lands in your repository, so ownership is total.
The highest strict pass rate in the agent group, and third highest in the whole index. It was the only agent that produced a signed webhook with a retry ladder in prompt 5 in all five runs without the criterion being restated, and the only one that added an idempotency key to the write path unprompted. It is a terminal agent with no interface of its own, so design and SEO reflect the stack it was pointed at rather than anything it invented.
Index gap
1.65
Claude Code leads
Axes won
3 / 6
Cursor versus Claude Code, of ten axes
Faster median
Cursor
53.4 s per prompt
Higher pass rate
Claude Code
91.1%
Every measurement, side by side
Cyan marks the better figure on each row. The delta column is the first product minus the second.
| Measurement | Cursor | Claude Code | Delta |
|---|---|---|---|
| Composite index | 78.65 | 80.30 | -1.65 |
| Agent performance (weight 18) | 92.0 | 95.0 | -3.00 |
| Reliability (weight 18) | 86.7 | 91.1 | -4.40 |
| Scalability (weight 13) | 82.0 | 86.0 | -4.00 |
| SEO and GEO (weight 12) | 58.0 | 62.0 | -4.00 |
| API and MCP (weight 9) | 74.0 | 78.0 | -4.00 |
| Integrations (weight 8) | 80.0 | 76.0 | +4.00 |
| Design output (weight 7) | 66.0 | 68.0 | -2.00 |
| Speed (weight 7) | 58.8 | 47.9 | +10.90 |
| Value (weight 5) | 82.0 | 80.0 | +2.00 |
| Code ownership (weight 3) | 99.0 | 99.0 | +0.00 |
| Median seconds per prompt | 53.4 | 65.5 | -12.10 |
| p10 seconds | 28.7 | 39.9 | -11.20 |
| p90 seconds | 82.7 | 100.7 | -18.00 |
| Executions failed | 3 | 1 | +2.00 |
| HTML bytes, no JS | 24.0 KB | 26.0 KB | -2048.00 |
| LCP milliseconds | 1,720 | 1,680 | +40.00 |
| CLS | 0.030 | 0.030 | +0.00 |
| Entry price EUR | 19.00 | 17.00 | +2.00 |
Screenshot diff
Both landing pages, captured by the lab on 14 Aug 2026 at 1440 by 900. Stored locally, never hotlinked.
Cursor, captured 14 Aug 2026.
Claude Code, captured 14 Aug 2026.
Speed spread on a shared scale
Cursor
28.7 / 53.4 / 82.7 seconds
Claude Code
39.9 / 65.5 / 100.7 seconds
Axis by axis, who wins
Cursor wins 3 of 10
- Integrations80.0
- Speed58.8
- Value82.0
Claude Code wins 6 of 10
- Agent performance95.0
- Reliability91.1
- Scalability86.0
- SEO and GEO62.0
- API and MCP78.0
- Design output68.0
Verdict
Claude Code takes the composite by 1.65 points. It wins on the weighted total, but the axis table is where the real decision sits: Cursor takes 3 axes and Claude Code takes 6.
If the work in front of you is mostly interface, weight the design and speed rows. If it is a product with an API, webhooks and background jobs, weight agent performance, reliability and the API and MCP row, which together carry 45 of the 100 index points. If the project has to outlive its vendor, read the code ownership row first and treat everything else as secondary.
Both products in this comparison were measured in the same cycle, on the same specification, by the same harness. The per run data for each is on its product page and in /api/runs.json.