Which model does each kind of work, what to try when it fails, and how much Liro may spend.
Change anything below and this shows what it would have cost.
Cost
$209
Chat reply starts in
0.6 s
$500 cap reached on
Never
Pick per task. Prices are for the last 30 days of that task alone.
Chat
Answers in the assistant
Summaries
Documents and threads
Code
Agents that write or fix code
Search index
Making documents searchable
Tried in this order. The first one that can do the task answers, and the run notes which one it was.
Swift 3
Fast and cheap · starts in 0.6 s
Deep 3
Careful reasoning · starts in 2.1 s
Craft 1
Tuned for code · starts in 1.3 s
Counted from the 1st of each month. Admins are told at 80%.
When it's reached
Work carries on, on Swift 3, until the month resets.
Each person, per month
One busy agent can't use the whole workspace's budget.