Skip to main content
Open Workspace → Spent to see what your model requests cost and how they performed. Choose 24H, 7D, 30D or 180D, filter by network, and click refresh to update the data. Spent with cost, request, latency and token totals beside the chart and activity

Compare cost and performance

Total Spent shows settled spending. Requests includes the success rate and share of streamed responses; Avg Latency measures the full request, while Avg TTFT measures time to the first token. Tokens separates input and output. Click the chart to expand it, then choose SPEND, REQUESTS, TOKENS or LATENCY. Hover over a point to read its values; the latency chart compares average, P50, P95 and P99 timings. Expanded spending chart with Spend, Requests, Tokens and Latency controls

Find the models behind the bill

Click the model table to expand it. Search by model or provider, and sort by COST, CALLS, TOKENS, LATENCY, SUCCESS, RECENT or TYPE. The P95 column is the time within which 95% of requests finished. Expanded Models table with search, cost, calls, tokens, P95, success and type Switch to BY TYPE to compare spending and usage across text, image, audio and other model types.

Inspect requests and settlements

Expand the activity panel. REQUESTS shows individual calls and their status, timing, token counts, operation and settled cost. Use the status filter to narrow the list.
Expanded request activity with token counts, operations and settled costs
Choose SETTLEMENTS for payment records, including the inference charge and platform fee. Load more requests and Load more settlements open older records; follow the transaction links where shown. See Session, payment and prices for spending approval and workflow fees, or Earn for your creator royalties.