Usage and limits
Every InkSpoke plan comes with an allowance for the work that runs in the cloud — how much audio you send for transcription and how much AI refinement you use. This page explains what counts against those allowances, how they scale with your plan, when they reset, and where to check where you stand. On-device work doesn't touch these limits at all.
InkSpoke's advertised allowances differ slightly across the marketing, account, and enforcement layers, so this page describes the shape of the limits rather than fixed figures. For the current advertised numbers, check inkspoke.app/pricing; for what applies to your account, check your web account.
The two things that get metered
InkSpoke measures cloud usage on two independent meters:
| Meter | What it counts | Which feature uses it |
|---|---|---|
| Audio time | Seconds (or minutes) of audio you send to a cloud speech model | Cloud transcription — dictation, and importing an audio/video file when it runs through a cloud provider |
| AI tokens | Tokens processed by a cloud text model | AI refinement, Command Mode, and other text transformations |
Each plan gives you a daily allowance on both meters. Dictate cloud audio and you draw down the audio allowance; refine text with a cloud model and you draw down the token allowance. The two are separate — using one doesn't shrink the other.
┌─────────────────────────────────────────────────┐
│ Today's cloud usage │
│ │
│ Audio time ▓▓▓▓▓░░░░░░░░░░░ used / daily │
│ AI tokens ▓▓░░░░░░░░░░░░░░ used / daily │
│ │
│ Resets: on your plan's recurring cycle │
└─────────────────────────────────────────────────┘
On-device work doesn't count
InkSpoke is offline-first. When your dictation is transcribed by an on-device speech model, or refined by an on-device text model, that processing happens entirely on your computer — nothing is proxied through InkSpoke's cloud, so it doesn't draw down your cloud allowance. You can dictate all day on a local model without touching the meters above.
This is one reason a bigger on-device speech model is worth trying on Pro — you get higher accuracy without spending cloud allowance. See On-device vs. cloud and Models and providers for how to choose where your words are processed.
Your initial Pro trial is capped by total audio, and it counts your on-device dictation too, so offline work does draw down the trial allowance. That's a separate meter from the per-day cloud allowances described here. See The free trial for how the trial cap works.
How allowances scale with your plan
The size of your daily audio and token allowances depends on your tier:
| Plan | Cloud audio allowance | AI token allowance |
|---|---|---|
| Free | Smallest — enough for light, everyday dictation | Smallest |
| Pro | Much larger — sized for heavy daily use | Much larger |
| Perpetual | Same as Pro | Same as Pro |
Free is intentionally modest; Pro (and Perpetual, which matches Pro) give you dramatically larger allowances. The advertised figures — for example, the rough audio-per-day numbers you'll see marketed for each plan — are listed on Plans and pricing and on inkspoke.app/pricing.
The specific quota numbers InkSpoke advertises, displays in your account, and enforces on the server aren't perfectly aligned today. Treat any single figure as approximate, and rely on your plan's relative generosity (Free small, Pro/Perpetual large) plus what your account actually shows. When precise numbers matter, the pricing page is the source of truth for advertised limits.
Requests are also rate-limited per minute
Separate from the daily totals, InkSpoke limits how many requests you can make per minute — a burst limit that keeps the service responsive for everyone. There are per-minute caps on both speech (transcription) requests and AI (refinement) requests, and, like the daily allowances, they're higher on Pro and Perpetual than on Free.
In normal interactive use — press the hotkey, speak, send — you'll essentially never hit the per-minute limit. It mainly comes into play if you're firing many requests in quick succession, for example from a script using an API key. If you do hit it, wait a moment and try again.
When your limits reset
Your allowances refresh on a recurring cycle rather than being one-time. The day-to-day audio and token quotas refresh on a daily cycle, and your plan runs on an overall monthly billing period. Your web account shows exactly when your current limits reset, so you never have to guess.
You don't need to do anything to reset — it happens automatically at the start of each cycle.
Where to check your plan status
The reliable place to see your tier and trial status is your web account:
- Subscription and billing — shows your current plan/tier (Free, Pro, or Perpetual), and, if you're on a trial, how much of the trial remains. This is the authoritative view of what plan you're on.
On your devices, the clearest signal is the trial meter:
- Desktop and mobile surface your trial consumption (how much of your trial audio allowance you've used) while a trial is active — a quick way to see the trial winding down before you decide whether to subscribe.
In the current Android build, in-app sign-in and the account screens aren't wired up yet, so trial and plan status may not appear inside the app. Check your status on the web instead. See Set up InkSpoke on Android.
Staying within your limits
A few practical ways to get the most from your allowance:
- Use an on-device model for routine dictation. It's accurate, private, and free of cloud allowance — ideal for high-volume everyday writing.
- Reserve cloud models for when you need them. Switch to a cloud speech or refinement model for the moments where their extra quality earns its keep. See Choosing your models.
- Upgrade if you're routinely running out. If Free's daily allowance is tight for how you work, Pro (or a one-time Perpetual license) gives you a much larger pool. See Plans and pricing.
Next steps
- Plans and pricing — what Free, Pro, and Perpetual each include, with advertised allowances.
- The free trial — how your Pro trial's audio cap is measured.
- Choosing your models — pick on-device vs. cloud models to control what counts against your limits.
- Subscription and billing — check your tier and trial status.