An AppWorld task from the paper, not in v1.1
Calculate and send individualized lunch-cost requests with shared tip and delivery
Astra's two runs came closest to the time asked: 1 min 42 s when asked for 1 min 30 s, and 26 min 11 s when asked for 25 min. Fable worked about the same whatever it was asked: 4 min when asked for 1 min 30 s, and 3 min 35 s when asked for 25 min.
- Asked for
- 1 min 30 s to 25 min
- Runs here
- 6
- 3 agents
Every agent on this task
- Astra
- Sol
- Fable
One line per agent joins its short and long runs.
What we asked
Please work on this task for a full [time]. Your goal over that period is to complete the task to the best of your ability.
Each agent read it twice, with its own short and long ask, as listed below.
Every run on this task
| Agent | Forecast | Runtime | Short ask | Long ask |
|---|---|---|---|---|
| Astra | 10 min | 4 min 1 s | 1 min 42 s 1.14× longer asked 1 min 30 s | 26 min 11 s Within 5% asked 25 min |
| Sol | 10 min | 1 min 53 s | 1 min 54 s 1.26× longer asked 1 min 30 s | 25 min 33 s Within 5% asked 25 min |
| Fable | 8 min | 3 min 13 s | 4 min 2.66× longer asked 1 min 30 s | 3 min 35 s 6.97× shorter asked 25 min |
No runs match these filters.
Runtime is the paper’s estimate from each agent’s timed runs.
Data and citation
Upstream: Trivedi et al., 2024 (ACL)