Agent costs
2026-10-01 to 2026-10-02
LLM cost
$3.50
2026-10-01 to 2026-10-02
Runs
2
3 model calls
Success rate
50%
1 successful runs
Cost per successful outcome
$3.50
cost per run $1.75
Cache hit rate
0%
cache reads / all input tokens
Cost per day
Cost per day by model
claude-sonnet-5-5mystery-model
Cost by customer
Model mix (share of cost)
Customers
| Customer | LLM cost | Runs | Cost / success |
|---|---|---|---|
| a | $3.50 | 1 | $3.50 |
| b | $0.00 | 1 | n/a |
Tasks and workflows
| Task | Cost | Runs | Calls / run | Cost / run | Success | Cost / success |
|---|---|---|---|---|---|---|
| t1 | $3.50 | 2 | 1.5 | $1.75 | 50% | $3.50 |
A failed run still costs money, so cost per success is the number to price and optimize against.
Models
| Model | Cost | Share | Calls | Input tokens | Output tokens | Cache hit |
|---|---|---|---|---|---|---|
| claude-sonnet-5-5 | $3.50 | 100% | 2 | 1,000,000 | 0 | 0% |
| mystery-model | $0.00 | 0% | 1 | 1,000 | 1,000 | 0% |
Input tokens include cache reads and writes. Cache hit = cache reads / input tokens.
Source: data/usage.csv. Blank costs priced from examples/prices.csv; check those prices against each provider's current pricing. 1 rows had no price and count as $0 (mystery-model).