CodeLynq shows what work was delivered and what it cost at several levels: per run, per project and for the whole organization.
Costs are shown as you pay them. On CodeLynq credit that is in euros - the same amounts that are deducted from your credit and that spend limits use. If your organization uses its own model key or its own runners, CodeLynq does not charge for runs, and costs show what your model provider charges you, in dollars.
Analytics (organization)
Click Analytics in the sidebar. Choose a period of 7, 30 or 90 days.
| Figure | What it means |
|---|---|
| pull requests opened | Tasks that reached a pull request in this period - open, merged or closed since. Not the same as done. |
| tasks done | Tasks whose pull requests are all merged (reported by your git host), or that someone marked done. |
| run success rate | Completed runs as a share of runs that finished (completed plus failed). |
| first time right | Shipped tasks that did not need a new run based on feedback. |
| spent, per pull request and per finished task | The total cost of runs, and that total divided by the pull requests opened and by the tasks done. The last one is what accepted work cost. |
| median run time | How long a typical run takes. |
| runs and iterations | The number of runs, and how many of them were new runs driven by CI results or reviewer feedback. |
Costs are in euros when runs are charged to your model credit, and in US dollars - as your AI provider prices them - when your organization uses its own key or runners. The cost cards on the billing page say which.
Below the figures you find a Runs per day chart and a table per project with its runs, shipped tasks and cost.
Model routing
The Model routing panel shows how CodeLynq chooses a model per task, if your organization has turned that on (organization Settings → General, Pick the model automatically per task). It shows the runs per model, how often each got it right the first time, what it cost, and an estimate of what was saved compared with running everything on the most capable model. When routing is off, the panel explains what it does and links to the setting.
Per project: Delivered
Open a project and go to the Delivered tab. For 7, 30 or 90 days you see the tasks shipped and merged, what was spent and the cost per shipped task, the number of runs, failed runs and iterations, and a list of shipped tasks with their cost, date and a link to the pull request.
Per run
- The Runs page in the sidebar lists recent runs of all projects with their status, duration and Cost.
- On a task, next to Trigger New Run, an estimate shows what a run typically costs in this project, based on earlier runs.
- If a run would cost more than your organization lets runs spend without asking, it waits for approval first and shows the estimate. See Spend approval.
For the organization: Usage & billing
Organization Settings → Usage & billing shows:
- Usage: your plan, and how much of it you have used this month - development runs, seats and projects.
- Model credit: the balance that covers what the AI models cost, with recent activity.
- Spend limits: the limits on what runs may spend.
- Run usage cost: what runs cost this month and in total, with the number of runs and tokens.
See Billing and credits for how these work.
Notifications
Under Security settings (the shield icon at the bottom of the sidebar) → Notifications you can turn on Usage alerts: an email when your organization approaches its monthly run limit.