Pricing
Pensyve Cloud is usage-based. Every operation is metered, every account gets a free monthly allowance, and you can start without a credit card. Teams that outgrow pay-as-you-go can move to the Business plan for pooled allowances, seats, and discounted overage.
Plans
| Pay-as-you-go | Business | Enterprise | |
|---|---|---|---|
| Base price | $0/month | $99/month | Custom |
| Included standard ops | 1,000 | 50,000 | Custom |
| Included multimodal ops | 50 | 5,000 | Custom |
| Included extraction ops | 25 | 1,000 | Custom |
| Seats | 1 | 5 included, +$15/mo per additional seat | Unlimited |
| Overage rates | Standard | 10% discount | Volume discounts |
| Support | Community | Priority | Dedicated account manager |
Enterprise adds SSO (OIDC/SAML), audit logging, VPC appliance deployment, a 99.9% SLA, and GDPR compliance reports — contact sales.
Operation Tiers
| Tier | Pay-as-you-go | Business overage | What Counts |
|---|---|---|---|
| Standard | $0.002/op | $0.0018/op | remember, recall, forget, consolidate — all text-based operations |
| Multimodal | $0.005/op | $0.0045/op | Image memory (Florence-2) and code memory (UniXcoder) — embedding + storage + retrieval |
| Extraction | $0.01/op | $0.009/op | LLM-powered fact, entity, and relation extraction from text |
Overage is only billed once you exceed your plan's included operations for the month. Allowances reset monthly: on the 1st of the month for the pay-as-you-go free tier, and with your billing cycle on Business.
Why Three Tiers?
Not all operations cost the same to run:
- Standard operations use GTE-Small embeddings (384d) and SQLite/Postgres queries — lightweight and fast.
- Multimodal operations run specialized ONNX models (Florence-2 at ~230 MB, UniXcoder at ~125 MB) — 5–10× more compute per operation.
- Extraction operations invoke LLM inference for each call — the highest compute cost in the system.
You only pay for what you use. An agent doing pure text recall never pays multimodal or extraction prices.
Examples
| Use Case | Monthly Usage | Pay-as-you-go | Business |
|---|---|---|---|
| Personal agent | 800 standard ops | $0.00 (under free tier) | — |
| Development team | 5,000 standard + 200 multimodal | $8.75 | — |
| Production agent | 50,000 standard + 2,000 multimodal + 500 extraction | $112.50 | $99.00 (all within included pools) |
| Scale-up pipeline | 150,000 standard + 10,000 multimodal + 2,000 extraction | $367.50 | $310.50 ($99 + discounted overage) |
Rule of thumb: once your pay-as-you-go bill approaches ~$99/month — or you need more than one seat — Business is the better deal.
Free Tier
Every pay-as-you-go account gets monthly free operations:
- 1,000 standard — enough for most personal and development use
- 50 multimodal — enough to try image and code memory
- 25 extraction — enough to experiment with LLM extraction
Free operations reset on the 1st of each month. No credit card required to get started.
Billing
Billing is handled by Stripe. On pay-as-you-go, add a payment method from the Billing page (a quick Stripe setup checkout) — usage beyond the free tier is then billed monthly. The Business plan starts with a subscription checkout at $99/month. Usage is metered in real time — check the Billing page for current usage and estimated costs.
Invoices are generated monthly. All prices are in USD. Cancel anytime from billing settings.
Open Source
The Pensyve engine is Apache 2.0 licensed. You can always run it yourself for free — the cloud is for teams that want managed infrastructure, multi-tenant isolation, and zero operational overhead.