Pricing that scales with your relationships
Transparent USD pricing. Every plan includes AI-powered compatibility, BaZi timing, and the KKAI assistant. Pay for conversations, not seats you don't use. See the arithmetic.
Individual
Business — for matchmaking agencies & relationship coaches
Quota & cost — what a plan actually buys
Every plan is metered in KKOS conversations: one conversation is one complete modelling session, from the assistant's greeting to the finished 11-dimension model. The left half of this table is what you pay; the right half is what that conversation costs us to serve.
| Plan | Monthly | Conversations / mo | Effective $ / conversation | Overage $ / conversation |
|---|---|---|---|---|
| Free | $0 | 30 | — | — |
| Plus | $9.90 | 300 | $0.0330 | $0.040 |
| Pro | $19.90 | 1,000 | $0.0199 | $0.025 |
| Team 5 seats | $49 | 2,000 shared | $0.0245 | $0.020 |
| Business 20 seats | $199 | 10,000 shared | $0.0199 | $0.015 |
| Enterprise | Custom | Custom | — | Custom |
What one conversation costs us
Metered exactly, not estimated. The pipeline prices its own usage per session with these rates:
| Component | Unit price | Measured per conversation | Cost |
|---|---|---|---|
| LLM input tokens | $0.15 / 1M | 1,157 | $0.000174 |
| LLM output tokens | $0.60 / 1M | 2,445 | $0.001467 |
| Speech-to-text | $0.00 / min | 9.2 s | $0.000000 |
| Text-to-speech | $0.00 / min | 27.1 s | $0.000000 |
| Typical session, metered | — | — | $0.0016 |
| Observed mean per session | — | 15 sessions on record | $0.0020 |
The two totals differ because averaging per-session spend is not the same as costing average usage: short sessions hit the same prompt overhead as long ones, which pulls the recorded mean above the typical figure. The observed range across those sessions is $0.001 – $0.005.
Gross margin per conversation
| Plan | Revenue / conversation | Metered cost | Gross margin |
|---|---|---|---|
| Plus | $0.0330 | $0.0020 | 94.0% |
| Pro | $0.0199 | $0.0020 | 90.0% |
| Team | $0.0245 | $0.0020 | 91.8% |
| Business | $0.0199 | $0.0020 | 90.0% |
Free, honestly costed
30 conversations × $0.0020 is about $0.06 of metered compute per free account per month. Free is a real testing allowance, not a subsidy we hide behind a paid tier.
Paying yearly
Annual billing drops the effective rate to $0.0275 (Plus), $0.0166 (Pro), $0.0204 (Team) and $0.0166 (Business) per conversation — about 17% below the monthly rate.
What is not in this counter
Expressive voice synthesis is billed by the upstream voice provider and is not yet folded into the meter above, so the number is a floor for voice sessions. Modelling quality — not token price — is the real cost driver.
usage = recorded per-session totals on this deployment, September 2026 · rates = configured metering unit prices · allowances = GET /api/v1/plans
You can check the arithmetic against your own account: GET /api/v1/me returns model, challenge and usage counts, and each stored model carries its own cost breakdown. Metric definitions are in the evaluation methodology; hard limits are under quotas.
The ladder, drawn instead of tabulated
The same figures the tables above print — what one conversation costs you at each tier, and what serving one costs us. Shape, not a column.
Bar heights are the same figures the tables above print, stated in µ$ (one millionth of a dollar) so the ratios between them are exact — the dollar amounts are on each chip. Effective $ per conversation: Plus $0.0330 · Pro $0.0199 · Team $0.0245 · Business $0.0199. Metered cost of one conversation: output tokens $0.001467 of the $0.0016 total. Allowances and unit prices are the ones served by GET /api/v1/plans.