Cursor, Codex, and Claude Code are being compared again
Cursor’s is being viewed more favorably than in recent months, but it is still being compared with Anthropic and OpenAI tools. The rough view is that may sit somewhere between Sonnet 4 and Sonnet 4.5 in quality.
That raises a practical question: if Sonnet can be used heavily on a cheaper Anthropic plan, Cursor needs a clear extra reason to be worth it. GPT 5.5 and Opus 4.8 are seen as strong models, but recent use has felt uneven because of changing answer quality, models, API connection trouble, and slow .
Anthropic is singled out as feeling especially slow, and Opus feels more limited than before. Cursor’s future is framed around an assumed Elon Musk and SpaceX , with the hope that more backing could lead to better Composer models if the team stays intact.
Key points
- is being discussed more positively, but still has to prove its value against Anthropic and OpenAI tools.
- The rough quality estimate puts between Sonnet 4 and Sonnet 4.5.
- GPT 5.5 and Opus 4.8 are viewed as strong, but users are frustrated by uneven quality and issues.
- Cheap high-volume Sonnet access makes Cursor’s extra value less obvious for some users.
- The optimism around Cursor depends partly on an assumed SpaceX and future Composer improvements.