FinOps
FinOps turns your traces into financial and performance insight: what your AI costs, where the spend goes, how fast it responds, and which cheaper or faster model you could switch to without losing quality.
In the product, FinOps lives in the Gauge AI section of the sidebar. Cost numbers come from each generation's usage and your model prices, so keep those current for trustworthy figures.
What's in FinOps
| Page | In the product | Answers |
|---|---|---|
| Cost & Latency Dashboards | Gauge AI → Cost Dashboard, Latency Dashboard | Where does spend and latency go, by model, user, trace name and over time? |
| AI Usage | Gauge AI → AI Usage | How many traces, observations and scores do we record? |
| Cost & Performance Metrics | Gauge AI → Cost Performance Metrics, Latency & Analytics | How do cost, tokens and latency line up with quality scores? |
| AI Optimizer | Gauge AI → AI Optimizer | Which cheaper or faster model could this agent use safely? |
| Model Replay | Part of an optimizer run | How are comparison outputs produced, and on which providers? |
| Evaluation & Credits | Part of an optimizer run | How is quality judged, and how many credits does a run use? |
| Recommendations | Gauge AI → AI Optimizer → Recommendations | When is a model recommended, and what are the projected savings? |
| Models & Pricing | Settings → Models | How is cost calculated, and how do I set my own prices? |
Per-user cost is under Gauge AI → Users (see Users), and per-agent cost projections are in the AI Agent Center. Budgets, spend limits and gateway spend analytics belong to Budget AI.
How cost is calculated
For each generation, cost is the cost your trace reports, or else the sum over usage types of units × price, using the matching model definition. For a typical chat model that is input tokens × input price + output tokens × output price. Trace cost is the sum of its generations, and every other view (per model, user, trace name, environment, day) rolls up from there.
Access
| Area | Built-in roles with access |
|---|---|
| Dashboards, AI Usage, metrics views and AI Optimizer results | Owner, Department Admin, Security & Compliance, Cost Analyst |
| Clone dashboards, create and delete optimizer runs | Owner, Department Admin, Cost Analyst |
| Edit the model price catalog and LLM connections | Owner |
Custom roles can grant the Gauge AI module, or its sub-modules AI Optimizer (FinOps), Cost & Latency dashboards and Users (per-user cost), separately. See Members & Roles.