Cursor Composer 2.5 ships May 18 — Opus 4.7 / GPT-5.5 parity at $0.50 input / $2.50 output per M tokens

Cursor released Composer 2.5 on May 18 — its own in-house coding model that benchmarks at parity with Claude Opus 4.7 and GPT-5.5 on SWE-bench Verified, at prices of $0.50 per million input tokens and $2.50 per million output. The release confirms Cursor as a vertically-integrated model builder, not just a tooling wrapper.

The pricing matters as much as the benchmark. At $2.50/M output, Composer 2.5 is roughly 1/6th the cost of Opus 4.7 for sustained agent workloads. Cursor enterprise customers running thousands of background agents now have a meaningful cost lever they didn't before.

The strategic question: does Cursor stay vertically integrated (their own model + their own IDE) or open up to third-party labs (route to Anthropic / OpenAI / Google as backends)? The Composer 2.5 release suggests they're betting on vertical integration. That makes their $50B-ish valuation trajectory more durable but more expensive to defend.

Cursor — Composer 2.5 → · Lushbinary — agent comparison 2026 →