Pricing details
The token-level breakdown behind the headline pricing. If you want to verify the math behind typed's per-tier quotas and how they compare to the reference Pro / Max plans, this is where it lives.
1. Headline pricing
Solo tiers
| Tier | Monthly | Annual | Reference-plan usage (estimated) |
|---|---|---|---|
| Starter | $10 | $100 | Lighter than a typical Pro plan -- on-ramp tier for evaluation |
| Pro | $20 | $200 | Reference Pro plan standard tier |
| Max | $100 | $1,000 | Reference Max 5x standard tier |
Annual is a 2-month discount on all three tiers. The reference-plan column is provided for migration evaluators who want to map existing subscription CLI usage onto typed tiers; the underlying methodology and token quotas are documented in sections 2-3 below.
Per-seat pricing (Team / Team Max)
Team is $25/seat/mo and Team Max is $100/seat/mo. Each is a per-seat subscription for shared work inside an org. Each seat carries the per-tier quota envelope of the matching solo tier -- Team is Pro-equivalent per seat, Team Max is Max-equivalent per seat.
| Tier | Per-seat monthly | Quota envelope per seat | Billing period |
|---|---|---|---|
| Team | $25 / seat | Same as Pro (63.75M input + 1.7M output / month) | Monthly only |
| Team Max | $100 / seat | Same as Max (318.75M input + 8.5M output / month) | Monthly only |
How the seat math works. Our payment processor's volume pricing scheme bills the whole subscription at per_seat_unit_price * seat_count. Concretely:
- 3 seats on Team -> billed at $25 x 3 = $75 / month. The org-wide aggregate quota is 191.25M input + 5.1M output / month, shared across the seats in the org -- one heavy member can spend more than 1/3 of the pool without anyone's individual usage being capped.
- 4 seats on Team Max -> billed at $100 x 4 = $400 / month. Aggregate quota is 1,275M input + 34M output / month, again shared org-wide.
Adding a seat mid-period prorates against the remaining days; removing a seat takes effect at the next renewal. The full per-seat price applies for the full period the seat is active. Top-up shapes ($5 / $10 / $20 / $50) and overage rates are the same on Team / Team Max as on the solo tiers.
Hosted codebase indexing surface. Team and Team Max include hosted codebase indexing as an additional surface. You upload a source archive via the dashboard or POST /v1/codebases/{id}/files, we chunk and embed it, and queries from any seat in the org can retrieve relevant chunks at request time via the x-typed-codebase: <id> header. Storage and processing are kept inside our us-west-2 AWS region: source archives sit in S3 (SSE-S3 at rest), and chunk embeddings sit in our pgvector index on AWS RDS in the same region. Access is org-scoped -- a codebase belonging to org A is invisible to a user in org B even with the codebase's UUID. On cancellation, codebases retain a 30-day grace that allows resubscription without re-upload; after the grace expires they hard-delete within 14 days. You can purge any codebase immediately at any time by emailing privacy@typed.cloud from the address on your account. Full terminology and cross-border-flow disclosure live in the privacy policy ยง3.3a.
Local context works everywhere -- every Anthropic-API-compatible client reads files on your machine, on any tier. Hosted indexing is the additional convenience for orgs that want a shared, server-side index across seats; it is not a missing feature on solo tiers.
Why no Team annual?
Per-seat annual is deliberately deferred for V1. Per-seat annual on team SKUs adds proration and mid-year-seat-add complexity without obvious customer pull. We will revisit if Team customers ask. Solo Starter / Pro / Max annual still ship at the prices in the table above.
2. Internal token quotas
These are the per-month quotas typed's quota service enforces. They are not displayed on the marketing pricing page (which uses reference-plan framing per our pricing policy), but they are the underlying mechanism and are documented here so migration evaluators can map their existing subscription CLI usage onto typed's tiers.
| Tier | Input tokens / month | Output tokens / month |
|---|---|---|
| Starter | 21.25M | 850K |
| Pro | 63.75M | 1.7M |
| Max | 318.75M | 8.5M |
Quotas reset at the start of each calendar month UTC. The dashboard usage page shows your current period's usage in real time.
Why two numbers per tier, not one
Input and output tokens have very different costs and very different shapes in a coding workflow. A typical AI coding CLI session reads a lot more code than it writes -- the realized ratio runs in the 20-40x range once prompt caching and tool-loop context are factored in. The quotas above are skewed strongly toward input to match that shape; the output quota covers normal coding-workflow output without binding before the input quota does. (Per tier, the input:output ratio is roughly 25:1 for Starter and 37.5:1 for Pro / Max, matching the 2026-05-30 workload-shape rebalance and the 2026-05-31 / 2026-06-05 margin trims.)
If your workflow has an unusual ratio (heavy code generation, multi-file scaffolding, large test-suite generation), top-ups add both input and output tokens to your remaining balance at a fixed per-dollar rate. The exact split is the same on every top-up; you do not need to choose a side at purchase.
3. Equivalence to reference plans
The reference Pro / Max plans do not officially publish per-tier token quotas -- the quotas are advertised in usage-window terms (5-hour rolling windows, weekly caps, etc.) rather than monthly token totals. So the equivalence below is estimated.
Methodology: we measure how many tokens a typical Pro / Max session typically consumes over the published usage windows, multiply out to a monthly figure, and pick typed quotas that match within an order-of-magnitude grain. The intent is that a customer who maxed out a typical Pro plan under typical coding load will not max out typed Pro under the same load.
Shape caveat. typed's quota is heavily skewed toward input tokens (roughly 37.5:1 input:output on Pro) because realized coding-workflow traffic runs at that ratio once prompt caching and tool-loop context are factored in. The tradeoff: typed quotas favor read-heavy workflows (refactoring, debugging, multi-file investigation) over output-heavy ones (large test-suite generation, multi-file scaffolding from scratch, heavy boilerplate emission). If your existing usage is dominated by output volume, your typed quota may feel tighter than the headline equivalence suggests; bench against your shape before committing.
The estimate is approximate, not certified. If you find that typed Pro feels tighter than a typical Pro plan on your specific workflow, email support@typed.cloud with a description of the workflow -- we collect this data to refine the quotas over time.
Cache eligibility stretches the numbers further. Prompt caching reduces effective input cost by roughly 50% on repeated-prefix workflows -- which is most of any AI coding CLI's traffic (system prompt + tool definitions + recently-referenced files get re-sent on every turn). The headline token quotas above are billed in raw tokens; the effective working capacity on cache-friendly workflows is meaningfully higher.
4. Overage rates
When you exceed your monthly quota, you can either upgrade tiers or buy a top-up. The top-up rate is the same on every paid tier.
| typed | Reference CLI (Extra Usage) | typed advantage | |
|---|---|---|---|
| Input | $1.67 / M tokens | $3.00 - $5.00 / M tokens | ~44% - 67% cheaper |
| Output | $8.33 / M tokens | $15.00 - $25.00 / M tokens | ~44% - 67% cheaper |
Top-ups are sold in fixed $5, $10, $20, and $50 denominations. Each top-up adds both input and output tokens at a fixed per-dollar split (handled server-side, not a customer choice at purchase). The dashboard shows your remaining input and output token balance directly. Top-ups expire 90 days from purchase and are consumed before the next month's quota refills.
| Top-up | Input tokens added | Output tokens added |
|---|---|---|
| $5 | 1.5M | 300K |
| $10 | 3M | 600K |
| $20 | 6M | 1.2M |
| $50 | 15M | 3M |
The price differential comes from two things: typed's underlying cost-per-token is lower than the reference CLI's published rates, and typed runs at a thinner top-up margin than the reference CLI's overage rates appear to imply. The margin still leaves enough headroom for sustainable operation; we are not subsidizing.
5. Auto-top-up at quota: optional, opt-in
Unlike subscription CLIs' 5-hour rolling windows and weekly caps, typed bills monthly. By default, requests past quota return HTTP 429 with a one-click top-up prompt -- pick a $5/$10/$20/$50 denomination and continue. Enable auto-top-up in your dashboard and the next request past quota auto-charges your configured denomination, emails you a receipt, and the session continues without prompting.
| Usage | Behavior |
|---|---|
| 0% - 80% | Normal. No banner, no friction. |
| 80% - 100% | In-app banner; email warning showing how many tokens remain. |
| 100% and beyond, auto-top-up off (default) | HTTP 429 with a one-click top-up prompt -- pick a denomination and continue. |
| 100% and beyond, auto-top-up on | Next request auto-charges your configured denomination; session continues; receipt emailed. |
Dashboard settings: toggle on/off, denomination, and an optional monthly ceiling that caps how much auto-top-up can charge in a calendar month.
6. Cache discount
Prompt caching is enabled on every paid tier and applied automatically -- you do not need to set a flag or use a non-standard API shape. If your client sends repeated prefixes (most coding clients do, because they re-send the system prompt and recent context on every turn), the repeated prefix portion is billed at roughly half the normal input rate.
We do not surface a separate "cached input" line in your usage breakdown today because the discount happens at the cost basis level, not at billing. You see the headline token count in the dashboard; the cache discount shows up as typed being able to serve the same workload within the same monthly quota and at the same overage rates as our cost basis improves.
7. Billing structure: monthly vs subscription CLIs' 5-hour windows
Subscription AI CLIs bill against rolling 5-hour usage windows and weekly caps. typed bills monthly with auto-top-up at quota (see section 5 above).
Pick the shape that fits your work pattern:
- Steady daily usage: typed's monthly billing is friendlier -- you do not lose unused capacity at the end of every 5-hour window.
- Bursty work (long sessions twice a week, idle the rest of the time): subscription CLIs' rolling windows may suit you -- the window resets after each idle stretch.
Most coding work is closer to "steady daily" than "bursty," which is why typed defaults to monthly. If your work pattern fits the subscription CLIs' windows better, keep using your existing CLI for that workload -- the two services interoperate via the same API shape.
8. Currency, taxes, and renewals
- All prices are in USD.
- VAT / GST / sales tax is added at checkout where applicable, based on your billing address. Our payment processor (Lemonsqueezy) acts as Merchant of Record and remits tax for you.
- Subscriptions renew automatically on the same day each month (monthly plans) or each year (annual plans).
- Cancellation takes effect at the next renewal -- the remainder of your paid period is served normally. See the migrate guide section 5 and terms of service for the full sales-final policy.
9. Open questions
If anything on this page is ambiguous, or if you want the methodology behind a specific number explained in more detail, email support@typed.cloud. We will answer and -- if the question is shared by other customers -- update this page so the answer is public.
Last updated 2026-06-05.