OSWorld 2.0, longest of three requests

GPT 5.6 Sol on “Invoices, photos, and email”

Shorter

10.49× shorter

Stopped at 10% of the time asked, 3 h 46 min short.

Asked

Worked

Safety cutoff 8 h 20 min, off the scale

Every detail of this run

Agent: GPT 5.6 Sol, in Codex
Benchmark: OSWorld 2.0
Task: Invoices, photos, and email
Asked for: 250 minutes: “Please work on this task for a full 250 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked: 23 min 50 s (1,430 s)
Ratio: 0.10× the request
How it ended: Ended on its own
Grade: Passed; It met every checkpoint. The grader returned 1. Pass rule: Binary completion: a score of 1.00, with every checkpoint met.
Score: 100 of 100; Graded during the run. The paper's value.
Wording: The standard sentence, used from 4 September 2026.
Run id: bs262ddupn
Data release: 26 Sep 2026

The other requests for GPT 5.6 Sol on this task, and the other agents at this request
Sol on this task	Asked	Worked	Result
Shortest	15 minutes	58 min 59 s	3.93× longer
Middle	60 minutes	1 h 3 min	Within 5%
Longest (this run)	250 minutes	23 min 50 s	10.49× shorter
Other agents, longest request
GPT 6 Astra	250 minutes	4 h 15 min	Within 5%
Claude Fable 5.1	250 minutes	39 min 40 s	6.30× shorter

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

Data and citation