Terminal-Bench 4.0, shortest of three requests
GPT 6 Astra on “Repair corrupted pretraining shards and reproduce the intended checkpoint”
Within 5%
Within 5%Worked 101% of the time asked, within 5% of it.
- Asked for
- 20 minutes
- Worked
- 20 min 18 s
- Grade
- Passed
- Score 100 of 100. Graded during the run.
- How it ended
- Ending not labelled yet
Every detail of this run
- Agent
- GPT 6 Astra, in Codex
- Benchmark
- Terminal-Bench 4.0
- Asked for
- 20 minutes: “Please work on this task for a full 20 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 20 min 18 s (1,218 s)
- Ratio
- 1.01× the request
- How it ended
- Ending not labelled yet
- Grade
- Passed
- It passed the task's tests. The grader returned 1. Pass rule: The task's tests pass.
- Score
- 100 of 100
- Graded during the run. The paper's value.
- Wording
- The standard sentence, used from 4 September 2026.
- Run id
- agsaetfkon
- Data release
- 26 Sep 2026
Related runs
| Astra on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Shortest (this run) | 20 minutes | 20 min 18 s | Within 5% | |
| Middle | 80 minutes | 1 h 21 min | Within 5% | |
| Longest | 300 minutes | 5 h 1 min | Within 5% | |
| Other agents, shortest request | ||||
| GPT 5.6 Sol | 20 minutes | 20 min 15 s | Within 5% | |
| Claude Fable 5.1 | 20 minutes | 33 min 46 s | 1.69× longer | |
See it on the chart. Transcripts are not published yet.
Read before quoting. How this run ended has not been labelled yet. It is timed like every other run and counted in every number. Details
Data and citation
Upstream: Merrill et al., 2026 (ICLR)