ClawsGO Docs

Credits

Credits are the unit of account on ClawsGO. Model usage is metered per token at each model's own rates; the multiplier on the menu is a reference ratio.

Credits are the only unit ClawsGO charges in; 3,000 credits correspond to one US dollar. Plans grant credits every month, top-ups and rewards arrive as credits, and model calls and per-use services in your tasks are deducted from them.

Monthly and permanent credits

Your balance has two parts:

  • Monthly credits: granted every 30 days from the day you bought the plan, independent of calendar months. On a monthly plan they are valid for that cycle and whatever is left expires at the end; on a yearly plan, leftovers roll into permanent credits.
  • Permanent credits: top-ups, signup credits, referral rewards and yearly rollovers. They never expire.

Monthly credits are spent first, then permanent credits. After a plan ends, permanent credits keep working with Free-plan models only.

How model usage is metered

cost = input × input rate + cached input × cached-input rate + cache writes × cache-write rate + output × output rate
  • Four rates: every model has its own rates for input, cached input, cache writes and output, and the gap between models differs by lane. Cached input costs far less than normal input. Caching is automatic; there is nothing to configure.
  • The reference multiplier at the end of each row in the model menu is roughly what the model costs on a typical task relative to ClawsGO 1.5 Silicon (x1.0), for example ClawsGO 1.5 Lithium x0.5, Claude Sonnet 5.5 x1.5, Claude Opus 5.5 x3.0. It is a guide only: actual charges follow the rates above, and the real ratio shifts with a task's token mix. The full table is in Choose a model.

The 512K and 1M context windows and Cluster mode carry no feature fee, but the first feeds the model more per turn and the second runs several subagents, so a task's total consumption rises. Subagent calls (Delegate and Cluster modes) are charged at the subagent model's rates, which default to the main model's.

Per-use services

ServiceCredits
Web search5 per search
Image generation100 per image, failed generations are free; GPT Image 2.5 Sunburst needs Pro or above
Vector generation200 per image, failed generations are free
Cloud hostFree

Failed or interrupted tasks

Charges follow usage that actually happened, not whether the task finished. If you stop or interrupt a task, what it already used is still deducted. If a task fails because of a fault on our side, open a ticket under Settings → Tickets and the credits are returned once confirmed.

Running out

If the balance is too low when you send, the message is not sent and "Not enough team credits" appears above the composer. If the balance runs out mid-task, an "Out of credits" card appears at the end of the reply; the work so far is safe. Top up or upgrade, then reply "Continue".

Team owners can set each member a credit limit per 30-day cycle. A member who reaches it sees "Your spending limit for this cycle has been reached" and is paused until the next cycle or a higher limit.

Viewing usage

Open Team & Subscription → Usage for every credit coming in and going out. Deductions for model calls, web search, image generation and the like show which member triggered them and the tokens used.