How to Check Cline Usage and Cost

Cline is one of the most popular open-source autonomous coding agents — a VS Code extension with roughly 62k stars on GitHub. Unlike a hosted subscription tool, Cline is bring-your-own-key: you plug in an Anthropic, OpenAI, OpenRouter, or other provider key, and Cline drives that model through plan-and-act loops. Because every request is billed to your key, knowing what a task actually costs matters before it runs away on a multi-file refactor. The good news is that Cline is honest about the number in real time. The catch is that the number is ephemeral.

How do I see Cline's token usage?

Cline surfaces usage directly in the task panel — you do not need a separate command or dashboard. While a task is running, the header of the conversation shows a compact meter that updates after every model turn:

1. 1

   Open the Cline panel in VS Code

   Click the Cline icon in the Activity Bar (or run the “Cline: Open In New Tab” command). The current task fills the panel.
2. 2

   Read the task header meter

   Near the top of the active task, Cline prints tokens up and tokens down (the arrows ↑ and ↓), the running cost in dollars, and a cache read/write count when prompt caching is in play.
3. 3

   Expand the API request rows

   Each “API Request” entry in the timeline can be expanded to show the tokens and cost for that single call, so you can trace which step was expensive.

The arrows are the key to reading it correctly. The up-arrow is the input (everything Cline sent the model: your prompt, the system prompt, file context, and the full prior transcript). The down-arrow is the output the model generated. On an agentic tool like Cline the up-arrow dominates, because each act step re-sends the growing context.

Where is my real Cline bill?

Because Cline is bring-your-own-key, the dollar figure in the sidebar is an estimateCline computes from the provider's published per-token prices. The authoritative charge is whatever your model provider records. To reconcile the two, open the usage or billing page of whichever key you configured:

1. 1

   Anthropic API key

   Check the Usage tab of the Anthropic Console — it breaks spend down by day and by model, which you can line up against Cline's per-task estimates.
2. 2

   OpenAI key

   The OpenAI platform billing dashboard shows daily usage; Cline's estimate should land close to it once you account for cached input pricing.
3. 3

   OpenRouter key

   OpenRouter's Activity page logs each request with the exact credits spent, which is the closest thing to a per-request receipt for Cline traffic.

Cline's estimate and the provider's invoice can diverge for a simple reason: prompt caching. When Cline re-sends context that the provider already cached, those cached input tokens are billed at a fraction of the normal input rate. A good mental model for what a task costs is:

The cached rate is typically about a tenth of the fresh input rate, which is why a long Cline task with heavy context reuse can cost far less than a naive tokens-times-price guess would suggest. Where most of a single task's tokens actually land usually looks like this:

Does Cline show total cost over time?

No — and this is the honest limitation worth planning around. Cline's meter is scoped to the current task. Start a new task and the count resets to zero. Cline keeps no rolling daily total, no weekly history, and no view that adds up the cost of all the tasks you ran this month. It also has no concept of your other tools — if you also drive Claude Code in the terminal or Codex CLI, Cline has no idea those exist.

So if you want to answer questions like “what did I spend on AI coding this week?” or “is my Cline habit bigger than my terminal-agent habit?”, the in-editor meter cannot get you there. You have two honest options: scrape your provider's billing export by hand, or use a tool that keeps a cross-time, cross-tool ledger for the agents that do write local logs.

How do I track Cline alongside my other agents?

Most people who run Cline also live in a terminal agent or two. That is exactly the gap whoburnedmore fills — not for Cline, but for the agents that leave readable usage logs on your machine. Running one command reads those logs and gives you the persistent rollup Cline cannot:

$ npx whoburnedmore↳ scanning local agent logs…   TOKENS BY TOOL · last 30 days  ────────────────────────────────────────  claude code      18.3M    est. $61.40  codex cli         7.9M    est. $24.10  gemini cli        4.2M    est. $ 9.80  opencode          2.6M    est. $ 7.30   (Cline is VS Code-native — read its sidebar for that)  cross-tool total:  33.0M tokens   est. $102.60

whoburnedmore reads the local logs that Claude Code, Codex, Gemini CLI, opencode, and more terminal agents already keep, then aggregates them into a daily and monthly breakdown with estimated cost — and ranks you on the leaderboard if you choose to submit. It is the complement to Cline's live meter: Cline tells you what the task in front of you costs right now, whoburnedmore tells you what your terminal agents added up to over time.

A simple two-tool workflow

In practice the cleanest setup is to let each tool do what it does best. Keep Cline's panel open while you work so the live meter warns you before an Act-mode loop gets expensive, then once a week glance at whoburnedmore for the longer arc across your terminal agents. The two together cover both the “right now” and the “over time” questions without pretending either tool does the other's job.

Want the same treatment for a tool whoburnedmore reads directly? See the Claude Code usage guide for a deep dive into reading those local logs.