How to Check Windsurf Credits and Cost

Windsurf is an AI-native IDE — a Visual Studio Code fork built around its Cascade agent — and one of the more prominent alternatives to Cursor. You can find the product at windsurf.com. In 2026 Windsurf shifted to a credit model, where each Cascade action draws down a pool of credits attached to your plan. The change made billing more predictable for the company but, for many users, more opaque: it is genuinely hard to know what a single edit, search, or tool call really cost you in credits before you run it.

How do I check my Windsurf credits?

Everything you need lives inside the editor and your account page — there is no command to run for this part. Windsurf intentionally keeps the meter close to where you spend.

1. 1

   Open the in-app credit meter

   Launch Windsurf and open Settings, then the Plans & Usage panel. The meter shows credits consumed against your plan's monthly allotment and how much of the cycle remains.
2. 2

   Check the per-action breakdown

   Cascade conversations list the actions taken — model calls, edits, terminal commands. This is the closest you get to seeing which work drained the most credits, though Windsurf does not always print an exact credit figure beside every single step.
3. 3

   Confirm against your account dashboard

   Sign in on windsurf.com and open the billing area. The web dashboard mirrors the in-app meter and is the authoritative number when you near a top-up or overage.

What does a Windsurf credit actually cost?

This is the question users find hardest, and the honest answer is: it depends on the model and the action, and Windsurf controls that mapping. A credit is an internal unit, not a token and not a dollar. Mapping credits to a real dollar figure means working backward from your plan:

Multiply that rate by the credits the meter says you burned this cycle and you have a rough dollar value of your Windsurf usage. We are deliberately not quoting Windsurf's credit prices here — they change, and the authoritative figures are the ones on your own Plans & Usage screen. The deeper problem remains: the credit unit hides the token spend underneath, so you cannot easily compare a Cascade edit to the same work done by a terminal agent that bills in raw tokens.

How do I track my other AI tools' spend too?

Most people who use Windsurf also keep a terminal open running agents that do write local usage logs. That is exactly the territory whoburnedmore covers — Claude Code, Codex, Gemini CLI, opencode, and the newer agents like Amp, Droid, and Goose. It never looks at Windsurf; it reads only the files those CLI tools already store on your machine.

$ npx whoburnedmore↳ scanning local CLI usage logs…   TOOL          TOKENS       EST. COST  claude code   11.6M        $38.90  codex          5.2M        $13.40  gemini cli     2.4M         $3.20  windsurf      (not read — see in-app meter)  ─────────────────────────────────────────  cli total     19.2M        $55.50

Is a credit model better or worse than per-token?

A credit pool is easier to budget against — you know your monthly ceiling up front, the same way Cursor's request model does. The downside is visibility: when the meter ticks down you cannot always see which action cost what, and you cannot diff Windsurf against a per-token agent without doing the conversion by hand. Terminal agents trade that predictability for transparency, exposing exact token counts you can price at published rates.

When the Windsurf credit model suits you

If you live inside the IDE and value a fixed monthly cap over line-item detail, the credit meter is enough. You glance at Plans & Usage, see how much of the cycle is left, and move on. For many developers that simplicity is the whole appeal.

When you want token-level truth

If you care about the exact cost of a task — or you run several agents and want them ranked head to head — the per-token view wins, and that is where whoburnedmore is built to help. For the IDE-versus-terminal billing distinction in more depth, the Cursor usage guidewalks through the same request-style abstraction with Cursor's numbers.

Does whoburnedmore touch my Windsurf account?

No — and it could not even if it tried. Windsurf stores credit usage server-side and exposes it only through the IDE and your windsurf.com login, so npx whoburnedmore has no Windsurf log to read, no Windsurf API key, and no access to your Windsurf account. It reads only the local files terminal agents keep — directories like ~/.claude and ~/.codex — and your prompts and source code never leave your machine. Run npx whoburnedmore --dry-run to preview exactly what it would submit before anything is sent. 🛡️

For the broader picture of what counts toward your AI coding bill across every tool, see the guide to checking AI token usage, which covers the full set of agents whoburnedmore reads and how the leaderboard ranks them.