Details
- Cursor announced that Claude Opus 5.5 is now available within its coding platform.
- The model records a 57.8% score on CursorBench at Max effort, according to Cursor’s post.
- Cursor describes Opus 5.5 as the top-performing model on its current CursorBench leaderboard.
- The company says Opus 5.5 costs 40% less per task than Claude Opus 5.
- CursorBench measures model performance on software-engineering tasks, while cost per task reflects the model’s token usage and published pricing.
- The release gives Cursor users access to Anthropic’s latest Opus model without requiring a separate workflow or platform.
- Cursor’s public benchmark materials caution that results can vary and that small score differences may not be statistically meaningful.
Impact
The announcement strengthens Cursor’s position as a model-agnostic coding platform by pairing a higher benchmark result with a claimed reduction in task cost. If the savings hold across users’ workloads, Opus 5.5 could encourage developers to shift more complex agentic coding tasks into Cursor. The competitive effect is likely to be incremental rather than decisive, since Cursor’s leaderboard includes several frontier models and benchmark results can vary by effort setting and task mix.
