Priced per analysis.
Access by conversation.
There are no tiers to pick and no seats to count. Visork bills one unit — a single image analysed. We are onboarding a limited number of pilot platforms, so access starts with a scoping call rather than a signup form.
One unit. No tiers to decode.
Content moderation costs scale with content, not with headcount — so the pricing does too. You send images, you pay for images. Everything the product does around that analysis, from the risk score to the audit log, is part of the same unit.
One analysis, one credit
The billable unit is a single image sent to the detection endpoint. The score, the verdict, and the audit log entry come with it — they are not metered separately.
No seats, no platform fee
You are not charged for team members, environments, or API keys. Two engineers or twenty, the invoice tracks one number: analyses processed.
Credits before commitment
Every pilot opens with 100 analyses at no cost, so you can measure detection quality against your own traffic before anything is signed.
Monthly volume
The dominant factor. Rates step down as throughput goes up.
Latency and throughput
Inline blocking at upload time carries different capacity requirements than an asynchronous review queue.
Retention and residency
EU processing is standard. Custom retention windows and isolation requirements are scoped per platform.
Support depth
Shared support and a named contact with response commitments are priced differently.
We are setting per-analysis pricing together with our first pilot platforms, against real inference cost and real volumes. Publishing a number now would mean publishing a guess. You will get a written quote before any integration work starts, and it will hold for the term we agree.
What exists, and what doesn't.
Visork is early, and a feature matrix full of asterisks would hide that rather than explain it. So here is the honest split: what runs in production today, and what is still being built. You should be able to decide from this list alone whether a pilot makes sense this quarter.
In the product today
LiveEvery item here ships at every volume. EU residency and the audit log are design decisions, not paid add-ons — there is no tier that unlocks them.
- Image-based CSAM detection
- REST API — one endpoint, JSON in, JSON out
- Probabilistic score on every analysis
- Threshold tunable per module
- Audit log entry for every call
- EU-hosted inference, storage, and logs
In the product today
LiveEvery item here ships at every volume. EU residency and the audit log are design decisions, not paid add-ons — there is no tier that unlocks them.
- Image-based CSAM detection
- REST API — one endpoint, JSON in, JSON out
- Probabilistic score on every analysis
- Threshold tunable per module
- Audit log entry for every call
- EU-hosted inference, storage, and logs
In development
Coming soonNot available yet, and not billable until it is. An uptime SLA lands as part of a commercial agreement rather than a self-serve toggle — pilot platforms shape the order the rest ships in. The Article 28 DPA is already published in full.
- Webhook notificationsComing soon
- Batch processing for archivesComing soon
- Compliance reporting exportsComing soon
- Uptime SLARoadmap
- Video, text, and multimodal modulesRoadmap
We don't publish the order these ship in. Pilot platforms hear the real sequencing under NDA during scoping.
Four steps to a number you can trust.
Because we quote against measured volume rather than a public rate card, the benchmark comes before the commercial conversation — not after it. Nothing is signed until you have seen how the model performs on your own traffic.
- 01Intro callThirty minutes. What you run, what you are processing, and what is forcing the decision now.
- 02ScopingWe agree the module, a threshold profile for your risk tolerance, and what a successful benchmark looks like.
- 03BenchmarkAPI key and 100 credits. You measure detection quality and latency against your own traffic, not our demo set.
- 04Commercial termsWe quote against the volume the benchmark actually showed. Nothing is signed before that number exists.
- 01Intro callThirty minutes. What you run, what you are processing, and what is forcing the decision now.
- 02ScopingWe agree the module, a threshold profile for your risk tolerance, and what a successful benchmark looks like.
- 03BenchmarkAPI key and 100 credits. You measure detection quality and latency against your own traffic, not our demo set.
- 04Commercial termsWe quote against the volume the benchmark actually showed. Nothing is signed before that number exists.
A CSAM detection endpoint is not something we hand to an anonymous signup. Every platform is verified before it gets a key — that is a property of the category, not a limitation of our stage.
We would rather say no early.
Pilot capacity is limited and a bad match wastes your quarter as much as ours. This is the honest read on who Visork can serve well right now — and who is better off talking to us in six months.
A good fit right now
- An EU platform processing user-generated images
- Volume that makes manual review impractical
- An engineering team that can ship a REST call
- A named owner for trust & safety or compliance
- Willingness to benchmark on real traffic
Probably too early
- You need video, text, or grooming detection today
- An uptime SLA is a precondition to piloting
- You need a published rate card to proceed internally
- There is no integration capacity for the next quarter
Tell us what you're processing.
Volume, content type, and the deadline you're working against are enough for us to come back with a scoped answer — including if the answer is not yet.