Cursor said Claude Fable 5.2 is now available inside its coding environment, and reported that the model ranks first on CursorBench at 60% when run at max effort.
Coding models · Cursor
Claude Fable 5.2 arrives in Cursor, with a reported lead on CursorBench
Fable 5.2 is now selectable in Cursor's coding environment. Cursor reports 56.9% on CursorBench at default effort, up from 49.2% for Fable 5.1, and 60% at max effort.
56.9%
Fable 5.2 at default effort on CursorBench
+7.7 pts
Gain over Fable 5.1 (49.2%) at default effort
60%
Fable 5.2 at max effort, reported as the top result tested
CursorBench score by model setting
All values in percent on the same scale. Column height is proportional to score.
49.2%
Fable 5.1
56.9%
Fable 5.2
Moving from default to max effort adds about 3.1 points in Cursor's reported figures.
Read with care
The figures are vendor-reported and leave key questions open.
• No independent third-party verification was included.
• No full task-level breakdown was published.
• Cost and latency at each effort level are not specified.
Positive reading
Fable 5.2 posts a clear gain over Fable 5.1 at default effort, and the max-effort setting offers a higher ceiling for complex code changes.
Critical reading
The results come from the company that built both the benchmark and the integration. Developers should test against their own workloads before judging the practical difference.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...