ProgramBench, AgentTime v1.1, asked for a quarter of its own forecast
Gemini 4 Argon on “facebook/zstd”
Shorter
1.54× shorterStopped at 65% of the time asked, 6 min 35 s short.
- Asked for
- 18.75 minutes
- A quarter of its own forecast of 1 h 15 min.
- Worked
- 12 min 10 s
- Grade
- Not graded
- How it ended
- Ended on its own
Every detail of this run
- Agent
- Gemini 4 Argon, in Antigravity CLI
- Benchmark
- ProgramBench
- Task
- facebook/zstd
- Asked for
- 18.75 minutes, a quarter of its own forecast of 1 h 15 min: “Please work on this task for a full 18.75 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 12 min 10 s (730 s)
- Ratio
- 0.65× the request
- How it ended
- Ended on its own
- Grade
- Not graded
- Clock tool
- The agent could call a wall-clock tool to read the current time.
- Wording
- The paper's standard sentence, sent in AgentTime v1.1.
- Run id
- vpyisgyb7q
- Data release
- 2 Nov 2026
Related runs
| Argon on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Quarter (this run) | 18.75 minutes | 12 min 10 s | 1.54× shorter | |
| Four times | 300 minutes | 41 min 40 s | 7.20× shorter | |
| Other agents, a quarter of their own forecast | ||||
| Claude Opus 5.5 | 45 minutes | 43 min 40 s | Within 5% | |
| GPT 6.1 Sol | 30 minutes | 31 min 11 s | Within 5% | |
| MiMo V2.6 Pro | 10 hours | 1 h 32 min | 6.55× shorter | |
| MiMo V2.6 Pro UltraSpeed | 8.75 hours | 52 min 41 s | 9.97× shorter | |
| Muse Wonder 1.0 | 4.25 hours | 5 min 29 s | 46.53× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Yang et al., 2026