Subscribe to edge-vms per camera, start a fixed-price safety pilot, or license any Dhi Labs product in minutes. Annual plans get two months free, and your price is locked for at least 24 months. Enterprise fleets are scoped pilot-first.
How pricing works
Every self-serve price on this page checks out instantly and is managed from a self-serve billing portal. The only custom number is the one that genuinely needs scoping: multi-site enterprise fleets.
Commercial model
Self-serve tiers have public prices and instant checkout. Enterprise deployments are scoped pilot-first.
Annual discount
Every annual plan is priced at ten months for twelve months of service, on every lane.
Price protection
Sign up now and your price is locked on that plan and quantity for at least 24 months.
Payment methods
Cards and wallets for monthly plans; ACH bank debit and USDC are available at checkout for larger purchases.
Billing unit
edge-vms bills per connected camera, because the camera is also the compute unit: one detection pipeline per camera.
edge-vms
Bring your own cameras. One detection pipeline runs per camera, so the camera is the billing unit.
2 cameras, 1 site
Try the core detection workflows on your own cameras.
per camera / month, billed annually
Full cloud video management with AI analytics.
per camera / month, billed annually, min 10 cameras
The full AI safety suite for warehouses and industrial sites.
invoice, pilot-first
On-prem and edge fleet architecture with SLA.
All self-serve paid subscriptions carry a $49/month account minimum. Manage plans, quantities, invoices, and cancellation any time at dhi-tech.com/billing.
Already using edge-vms? Subscribe from inside the app instead (Settings, Billing) and the plan activates on your organization instantly: vms.dhi-tech.com.
30 days, one workflow, one zone, up to 4 cameras
Prefer to scope it first? Talk to us
invoice-first, scoped after a validated pilot
Dhi Labs licenses
Research use is free under each repository's noncommercial license. Commercial use starts at $1,900/year per product: one company, up to 5 devices, license key issued instantly at checkout. Site licenses cover up to 50 devices with priority support.
per year, all 10 products
| Product | Developer ($1,900/yr) | Site (annual) |
|---|---|---|
| edge-scene-graphs | Buy license | $11,900 |
| thermal-perception | Buy license | $9,900 |
| face-free-reid | Buy license | $14,900 |
| Fleetmind | Buy license | $24,900 |
| fixed-camera-3d | Buy license | $12,900 |
| amodal-counting | Buy license | $7,900 |
| amodal-tracking | Buy license | $4,900 |
| Predictive Safety | Buy license | $39,900 |
| COVE | Buy license | $2,900 |
| prompt2model | Buy license | $29,900 |
One Developer license per company per product, hard cap 5 devices; 6+ devices needs Site. COVE is included free with Fleetmind and Predictive Safety Site licenses. OEM and redistribution licensing is quoted separately via contact.
20 model builds per month with commercial rights to trained models.
API access, fleet export, priority build queue.
Hosted inference APIs are usage-billed: standard capability at $2 per 1,000 calls, specialized capability (scene graphs, re-ID embeddings, thermal, metric-3D, calibrated counting) at $8 per 1,000 calls, with volume discounts above 100k calls/month and 1,000 free specialized credits monthly. API billing opens with the developer console; contact us for early access.
Self-serve prices are flat and published. Enterprise fleets price on deployment reality: where the cameras sit, which models run, how alerts route, and how much governance the site requires.
Enterprise pricing starts with the camera views that actually cover the risk zones, not the total camera count on a site map.
Compute depends on stream count, model load, hardware placement, and whether the node sits close enough to the camera network.
Forklift conflict, falls, smoke, crowding, track trespass, and restricted-zone events each have different tuning and validation work.
A simple local alert is different from Genetec or Milestone alarm routing with priorities, bookmarks, clips, and audit trails.
The cost model changes when events need horns, light stacks, supervisor tablets, dispatch workflows, or multiple recipients.
Retention, access control, clip approval, and security-review requirements affect implementation planning.
Each connected camera runs its own detection pipeline, so the camera is both the compute unit and the billing unit. You pick the quantity at checkout and can change it any time from the billing portal. Safety has a 10-camera minimum; overage bills at your tier's per-camera rate.
Two months. Every annual plan is priced at ten months for twelve months of service, on edge-vms, Labs licenses, and prompt2model hosted alike.
Sign up now and your price is locked on that plan and quantity for a minimum of 24 months. Expansions and upsells bill at the then-current list price.
Cards, Apple Pay, Google Pay, and Link on all plans; ACH bank debit and USDC stablecoin are available at checkout. Larger B2B deals can be invoiced with bank transfer terms instead: ask via the contact page.
When you need on-prem or edge fleet architecture, custom retention, an SLA, or a multi-site rollout. That path starts with a Focused Pilot so pricing is grounded in a validated deployment, not a site map.
Developer covers one company, up to 5 devices, with the license key issued instantly at checkout. Site covers up to 50 devices or cameras with priority support. OEM and redistribution licensing is quoted separately.
Use the checklist to define camera scope, alert path, success criteria, and rollout readiness before a pilot.
See the clearest first commercial wedge: forklift conflict, near misses, smoke, falls, and restricted-zone events.
Use the architecture comparison when pricing is tied to cloud streaming versus local edge inference.
Review camera ingest, edge compute, event routing, security posture, and deployment model before scoping.
Check Genetec, Milestone, Axis, NVIDIA Jetson, RTSP, and ONVIF fit before subscribing.
A buyer guide for comparing latency, camera fit, VMS integration, privacy posture, and pilot proof.
Bring the first risk zone, current camera environment, VMS constraints, and the operational outcome you want to measure. We will map that onto the right plan or pilot.
Best starting point: one zone, one incident class, one response owner.
See the flow on a real operating scenario and scope a pilot around one facility or corridor.
Review camera ingest, edge inference, alert routing, and what stays on-premises.
Download the deployment checklist buyers use before green-lighting an industrial AI pilot.
Bring camera count, VMS constraints, latency expectations, and privacy requirements to a technical review.