Pricing
Four tiers, one real rate table. Start free, self-hosted — pay only for the shared intelligence and governance layered on top as you scale.
This page is a static summary for reference. For the live, always-current numbers, see vouchstone.ai/pricing or run the free pre-assessment.
The four tiers
Open — $0
Free, self-hosted. The Vouchstone Open SDK and standalone harness runtime, yours to read line by line. Bring your own LLM keys — no procurement needed to start. Every new account also receives 5,000 welcome Vouchstone tokens (worth $50 of hosted-platform usage) to explore the rest of the platform.
Platform — $59/seat/mo + metered overage
The hosted Control Plane self-serve tier: extraction pipeline, Knowledge Graph wizard, Wiki, Company Brain + Council, default governance pack, and basic evals. Billed $59 per seat, monthly, with a 50-seat minimum — each seat includes 2,000 Vouchstone tokens/month, and usage beyond that meters against the real rate table below (1 USD = 100 Vouchstone tokens).
Enterprise — $59/seat/mo + $15,000/agent/yr
Everything in Platform, plus the governance infrastructure a compliance-gated, multi-team rollout needs: custom Constitution/Rule Book authoring (versioned Articles/Statutes beyond the seeded default pack) with per-action allow/deny and approval requirements, full autonomy-level gating (L0–L4, including unsupervised action), connector grants, a dedicated Lean Agent Gateway budget & rate-limit configuration, a hash-chained signed audit ledger export, and dedicated support. Priced as the same per-seat price as Platform (100-seat minimum), plus a flat $15,000/year governance fee per agent in production.
Sovereign & Air-Gapped — Contact us
For government, defense, financial services, and EU-sovereignty deployments that require a fully self-hosted, air-gapped installation — including model inference that never leaves your infrastructure. Sold as a premium annual license, scoped and quoted per engagement. Contact our sales team for pricing.
Token system (Open, Platform & Enterprise tiers)
Vouchstone uses a token-based usage model for hosted, metered usage:
- 1 USD = 100 Vouchstone tokens
- Tokens are deducted based on platform usage — see the rate table below
- Platform/Enterprise seats include a monthly token allowance; usage beyond the included allowance draws from your purchased token balance below. It's specifically the purchased balance that never expires — the monthly seat allowance resets each billing period.
- Purchased tokens never expire — they remain in your account until used
- 5,000 welcome tokens on registration
Usage rate table
Tokens are consumed based on the operations performed. This is the real rate table the platform bills against — the live version is served at /pricing.
| Operation | Cost (tokens) |
|---|---|
| Agent execution — simple chat | 15 |
| Agent execution — complex (tools/RAG) | 75 |
| Agent execution — advanced multi-step | 150 |
| Workflow run — base | 25 |
| Workflow — additional step | 10 |
| Workflow run — complex (conditionals) | 100 |
| Collective (multi-agent) run — base | 75 |
| Collective — additional agent | 25 |
| Collective — consensus mechanism | 50 |
| Memory query (semantic search) | 5 |
| Memory store — per 100 vectors | 10 |
| Memory store — per 1K vectors | 50 |
| Memory — advanced vector search | 8 |
| Document ingested | 20 |
| Document chunk created | 5 |
| Document indexed | 15 |
| LLM input — per 1K tokens | 2 |
| LLM output — per 1K tokens | 6 |
| LLM embedding — per 1K tokens | 1 |
| Decision graph traversal | 30 |
| Decision graph — node evaluated | 5 |
| Decision graph — conditional branch | 8 |
| Integration sync (Slack, JIRA, etc.) | 10 |
| Integration event received | 5 |
| Webhook trigger | 3 |
| External API call | 2 |
| Optimization Studio — DSPy run | 250 |
| Fine-tuning job | 1,000 |
| Fine-tuning job — large model | 2,000 |
| A/B test variant | 50 |
| Storage — per GB/month | 25 |
| Bandwidth — per GB | 10 |
| Agent / workflow create or update | Free |
Enterprise governance (included at Enterprise tier)
- Custom Constitution / Rule Book authoring — versioned Articles/Statutes beyond the seeded default pack, plus a per-engagement Authority Matrix (allow/deny, requires-approval, requires-dual-signoff per action kind)
- Autonomy levels (L0–L4) — from observe-only through fully autonomous, gating agent-originated actions
- Connector grants — an agent must be explicitly granted a connector before it can act through it
- Dedicated Lean Agent Gateway budget — a per-tenant/per-agent monthly USD budget cap plus per-minute rate limiting on agent actions, beyond the shared default
- Signed audit ledger export — every allow/deny decision hash-chained and signed for replay
- Dedicated support
- Priced at $15,000 per production agent, per year, on top of the same per-seat price as Platform
FAQ
Do credits expire?
No, credits never expire. They remain in your account until used.
What happens when I run out of credits?
Your agents will pause processing. You'll receive notifications at 20%, 10%, and 5% remaining credits.
Is there a minimum seat count?
Platform requires a minimum of 50 seats; Enterprise requires a minimum of 100 seats. Open has no seat minimum — it's unlimited-seat, self-hosted, for exactly this reason.
How is the Sovereign & Air-Gapped tier priced?
Sovereign & Air-Gapped is quoted per engagement based on deployment scope — it isn't published on this page. Contact sales for a quote.
Which tier is right for me?
Run the free, no-account pre-assessment — answer four questions and get a computed tier recommendation, rollout timeline, and price estimate with a full itemized "show your work" calculation trail.