Total tokens 01
Range
View
API-equivalent cost 02
Cache hit share 03
Model calls 04
Token flow
Uncached inputCache readCache writeOutput
API-equivalent cost
Public list rates · known-cost subtotal
By tool
Range and model filters applied. Select a row to focus the dashboard.
| Tool | Model calls | Uncached in | Cache read | Cache write | Output | Total | Cache hit | API cost | Cost ÷ 35 |
|---|
By model
Range and tool filters applied. Select a row to isolate a model.
| Model | Tool | Model calls | Uncached in | Cache read | Output | Total | Price | API cost |
|---|
Coverage & gaps
What the local files can—and cannot—support.
Reading the numbers
Methodology before interpretation.
- Tokens are disjoint buckets. Total = uncached input + cache reads + cache writes + output. Reasoning is already inside output and is never added twice.
- Costs are estimates, not charges. Standard public API list rates are used; subscriptions, credits, quotas, discounts, tools, taxes, and regional uplifts are excluded.
- DeepSeek uses peak rates. Grok uses below-200K tiers and Claude Haiku 5.5 uses its ≤100K tier. Actual bills can differ.
- “Unpriced” means unknown. Unpriced token volume is excluded from cost subtotals and flagged with an asterisk; it is never converted to $0.
- Dates are local to this box. America/New_York; weeks start Monday. The current day is partial.
Subscription-converted cost = API cost ÷ 35.
This is the owner’s comparison convention, not a subscription invoice or actual allocation.
This is the owner’s comparison convention, not a subscription invoice or actual allocation.
Price table & official sources
Price table & official sources
| Tool / local model | API pricing model | Input / 1M | Cached / 1M | Output / 1M | Cache writes / 1M | Status | Source |
|---|