EL

12d ago

Claude Sonnet (Sonnet 4.5) loses track of earlier instructions in long sessions. About 40-50 turns into a session, Claude Sonnet (Sonnet 4.5) starts forgetting constraints I set at the start — like the coding style guide or which files not to touch. Have to keep re-pasting instructions.

5

PA

Paula G.

11d ago

Claude Sonnet Cursor

Claude Sonnet (Sonnet 4.5) on greenfield vs existing repos: night and day. On a brand new project it's fantastic — fast, clean architecture. Point it at our five-year-old repo with inconsistent patterns and it struggles to match our conventions.

3

JA

Jamie L.

1mo ago

GPT-5 Cursor

Reproducible bug: Cursor mangles multi-file edits on GPT-5 (GPT-5.1). When a single instruction touches 3+ files, about 1 in 5 times one of the edits gets applied to the wrong file. Easy to catch in review but still concerning for anyone auto-applying diffs.

5

SA

Sam H.

1mo ago

GPT-5 Cursor

Inconsistent output for the identical prompt on GPT-5 (GPT-5.1). Ran the exact same prompt twice in a fresh session and got meaningfully different code structure both times — expected some variance, not this much.

3

TO

Tomasz W.

6d ago

Claude Sonnet Cursor

Reproducible bug: Cursor mangles multi-file edits on Claude Sonnet (Sonnet 4.5). When a single instruction touches 3+ files, about 1 in 5 times one of the edits gets applied to the wrong file. Easy to catch in review but still concerning for anyone auto-applying diffs.

5

RE

Ren T.

16d ago

GPT-5 Cursor

Went back to the previous version of GPT-5 (GPT-5.1) after trying the new one. The new version is smarter in isolated benchmarks but in my actual workflow it second-guesses itself more and takes longer to converge on a fix.

4

AI

Aino M.

21d ago

Claude Sonnet Cursor

Is Cursor Pro worth it over the base tier? Trying to figure out if the jump to Cursor Pro pays for itself. For heavy daily use it seems to, for occasional use the base tier is plenty.

1

J.

J. Alvarez

5d ago

GPT-5 Cursor

GPT-5 (GPT-5.1) is genuinely great at refactoring large files now. Asked it to split a 2,000 line controller into smaller services and it did it cleanly, preserved behavior, and even flagged an existing bug it found along the way.

2

DE

Dee C.

5d ago

Claude Sonnet Cursor

Claude Sonnet (Sonnet 4.5) feels noticeably faster since the update. Response latency on straightforward requests dropped a lot. Multi-file edits that used to take 20+ seconds now feel closer to 8-10.

4

MA

Marcus D.

12d ago

GPT-5 Cursor

Overall quality bump with GPT-5 (GPT-5.1) is real. Not hype — day to day the suggestions feel more aligned with what I actually meant, fewer 'technically correct but useless' answers.

4

RE

Ren T.

4d ago

Claude Sonnet Cursor

Reasoning quality on Claude Sonnet (Sonnet 4.5) varies a lot by task type. Excellent on algorithmic problems, noticeably weaker on ambiguous product-requirement style questions where it needs to make a judgment call.

5

MA

Marcus D.

1mo ago

Claude Sonnet Cursor

Claude Sonnet (Sonnet 4.5) feels noticeably faster since the update. Response latency on straightforward requests dropped a lot. Multi-file edits that used to take 20+ seconds now feel closer to 8-10.

3

OM

Omar S.

6d ago

GPT-5 Cursor

Cursor Pro limits: fine for solo work, tight for pairing. For my day-to-day solo coding Cursor Pro is plenty. The moment I try to use it for a full pairing session with a teammate watching, we blow through it in under two hours.

4

JA

Jamie L.

26d ago

GPT-5 Cursor

Cursor Pro price increase discussion. Not thrilled about the price change, especially since the usage caps didn't move proportionally. Curious how others are reacting.

3

Cursor

Community Summary

Topic Rooms for Cursor

Availability

Bugs

Had a different experience with Cursor?

Field Reports & Consensus

External Community (Reddit snapshot)

Coding

Context

Instruction following

Limits

Model comparison

Pricing

Quality

Reasoning

Reliability

Speed

Subscription

Tool use

Activity Pulse