ProgramBench, middle of three requests

Claude Fable 5.1 on “mookid/diffr”

Shorter

1.11× shorter

Stopped at 90% of the time asked, 5 min 43 s short.

Asked for
60 minutes
Worked
54 min 17 s
Grade
Did not pass
Score 92.4 of 100. Graded afterwards.
How it ended
Ended on its own
Every detail of this run
Agent
Claude Fable 5.1, in Claude Code
Benchmark
ProgramBench
Asked for
60 minutes: “Please work on this task for a full 60 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
54 min 17 s (3,257 s)
Ratio
0.90× the request
How it ended
Ended on its own
Grade
Did not pass
It did not pass every hidden test. The grader returned 0.924. Pass rule: Resolved: every hidden test passes.
Score
92.4 of 100
Graded afterwards. The paper's value.
Wording
The standard sentence, used from 4 September 2026.
Run id
k5fuvdlqss
Data release
26 Sep 2026
The other requests for Claude Fable 5.1 on this task, and the other agents at this request
Fable on this taskAskedWorkedSplitResult
Shortest15 minutes8 h 20 min33.33× longer
Middle (this run)60 minutes54 min 17 s1.11× shorter
Longest250 minutes2 h 16 min1.84× shorter
Other agents, middle request
GPT 6 Astra60 minutes1 h 1 minWithin 5%
GPT 5.6 Sol60 minutes1 h 1 minWithin 5%

See it on the chart. Transcripts are not published yet.

Data and citation

Upstream: Yang et al., 2026