How the credit wallet works

Your wallet is denominated in real dollars with an always-visible live meter, so you can see what you are spending as you spend it. Each model shows its $/message rate before you send.

Fast vs premium

• Fast models (for example Mistral Small or DeepSeek V4 Flash) are very cheap and effectively unlimited under fair use.
• Premium models (for example Claude Sonnet 4.6, GPT-5.5, Gemini 3.5 Flash) cost more — a premium message debits roughly 40× a fast message — so they draw from a paced premium budget.

How the premium budget is paced

• About 250 premium messages a month, released roughly 65 per week (25% weekly pacing) so you never run dry.
• Unused premium rolls over for about a week (with a 30-day ceiling).

When premium runs low

• At 80% of the week's premium, you get an announced warning banner — never a silent change. If you have enabled it, knecty can auto-switch you to a fast model (you can set this default in Settings → Models).
• At $0 premium, knecty falls back to a fast model so you can keep chatting — it is never a hard stop. You can top up for more premium right away.

Deep Research and other agentic work are metered in a separate token pool, so they never quietly eat your premium messages.