Story · Snorkel AI

Opus 5.5 vs Opus 5 vs Fable 5.1: Coding Benchmark Results (Snorkel AI)

blog post · Story page

Pass@1 is flat across three generations on one expert-created terminal-bench style task set: 61.5% for Fable 5.1, 60.7% for Opus 5, and 60.7% for Opus 5.5. Pass@5 separates them, and Opus 5.5's 76.7% sits below Opus 5's 79.3%.

In plain words

  • Snorkel found that newer coding assistants did not consistently do better on its programming tests.
  • The tests checked success on the first attempt and whether allowing five attempts helped.
  • Some Fable 5.1 failures came from stopping early or being unable to recover after errors.
  • People choosing coding help can use these failure details to compare assistants beyond their overall success rates.

Appeared in

Subscribe

Get the brief in your inbox

Pick daily, weekly, or both. Nothing is gated either way: every issue is on the site and in the feeds.

  • Weekdays at 8:45am IST, one lead story and 6 to 9 items.
  • Sundays, an argued synthesis rather than a recap.
  • One click to leave, and quiet days say so in the subject line.
How often

Weekdays 8:45am IST + Sundays. Unsubscribe in one click.

You're asking for The Agentic Brief by email at the cadence you picked. You can unsubscribe in one click from any issue, and your address is never sold or shared.