# Vynce plan reference

These are proposed monthly launch plans in USD. They are not subscriptions currently available for purchase. The website does not implement checkout, billing, usage metering, or paid agent access.

| Plan | Proposed monthly price | Seats | Included voice session minutes | Included task credits |
| --- | ---: | ---: | ---: | ---: |
| Personal | $29 | 1 | 60 | 200 |
| Pro | $79 | 1 | 240 | 600 |
| Teams | $199 | 3 | 600 pooled | 1,500 pooled |
| Enterprise | Custom | To be agreed | To be agreed | To be agreed |

Teams allowances are shared across all three seats, not multiplied by three. No plan promises unlimited use, continuous listening, immediate access, or a task's completion in one credit.

Personal targets individual voice, writing, coding, approved memory, and selected connections. Pro targets longer tasks and advanced computer workflows. Teams proposes shared workflows, workspace administration, and team activity visibility. Enterprise scopes larger deployments and custom integrations. Foundational approval, disconnection, activity, and memory controls are planned for every tier.

## Proposed accounting

Voice allowance measures open session duration, including silence and backend waits. The proposed GPT-Live 1 voice cost and Claude task cost are additive. Idle-session closure is required.

One task credit is proposed to permit up to $0.03 of metered backend AI spend. Each task uses at least one credit; total task spend rounds up in $0.03 units. Complex tasks, large context, retries, screenshots, and multi-step execution may require multiple credits. Heavy tasks should show an estimated range and maximum before they begin.

The application must reserve budget before each step, reconcile actual usage, release unused reservations, and stop safely at the allowance. These controls are design requirements, not implemented billing. No automatic overage charge is promised.

The proposal assumes standard first-party API costs, not promotional credits or an existing personal Claude subscription. Representative-workload testing is required before accepting payment.

## Provisional operating-cost forecast

These are internal monthly budget assumptions per account at full voice/task allowances, not vendor quotes or implemented billing. Teams includes three seats sharing the budgets.

| Component | Personal $29 | Pro $79 | Teams $199 |
| --- | ---: | ---: | ---: |
| Voice reserve ($0.055/minute) | $3.30 | $13.20 | $33.00 |
| Backend AI reserve ($0.03/credit) | $6.00 | $18.00 | $45.00 |
| Pooled hosting, database, storage, monitoring — assumed allocation | $2.00 | $4.00 | $12.00 |
| Capped executor runtime and paid tools — assumed budget | $1.00 | $4.00 | $12.00 |
| Payments — assumed 3% + $0.30 | $1.17 | $2.67 | $6.27 |
| **Forecast total** | **$13.47** | **$41.87** | **$108.27** |
| **Revenue left before other costs** | **$15.53** | **$37.13** | **$90.73** |

The pooled infrastructure allocation must cover actual provider-plan minimums, storage, egress, and concurrency at the intended account count. It does not provide a dedicated always-on machine per customer. Payment processing is a placeholder based on [Stripe's US domestic-card pricing](https://stripe.com/pricing) of 2.9% + $0.30, rounded up to 3%; subscriptions, international cards, conversion, disputes, and other fees can add costs. The residual excludes support, development, tax, refunds, and acquisition; it is not net profit.

[E2B's published compute rates](https://e2b.dev/pricing) provide an illustrative executor benchmark: $0.000014/vCPU-second + $0.0000045/GiB-second. A default 2-vCPU/4-GiB sandbox costs $0.00276/minute ($0.1656/hour) in variable compute. A required provider plan adds fixed cost (E2B Pro is $150/month plus usage) and must be allocated separately; no promotional credit is assumed. This is not a selected provider or verified desktop implementation.

Proposed supplementary execution limits are 120 / 480 / 1,440 aggregate sandbox minutes monthly, respectively, at the modeled resource ceiling; Teams shares its limit. Benchmark compute cost is about $0.33 / $1.32 / $3.97, leaving headroom in the $1 / $4 / $12 execution-dollar budgets. Enforce the time limit and dollar limit, whichever occurs first, plus one executing job per seat, ten minutes maximum per job, bounded tool steps, and idle shutdown. Paid tools consume the same budget; screenshot reasoning and other model calls still use task credits. These are provisional implementation controls, not an additional service currently sold. Stop at the budget; no automatic charges or unlimited computer use.

The existing prices and voice/task allowances remain provisional pending measured full-allowance and peak-concurrency costs.

See [the pricing page](/pricing) and [the proposed architecture](/reference/architecture.md).