OwnAI / Pricing
All prices INR · pre-GST · 18% GST applies

Transparent pricing. No per-token surprise.

One-time setup. Hardware you own. Annual maintenance optional. AI forever.

MODELOne-time + AMC PER-SEAT FEESNone PER-TOKEN FEESNone HARDWAREYours
§2 · Tiers

Three engagements. One pricing model.

Pilot · Production · Enterprise. Each tier is a one-time engagement with a fixed delivery scope, signed SOW, and named deliverables.

Pilot · 4 weeks

Prove the value before committing.

Discovery + a working PoC on a sample of your data, evaluated against a rubric you sign before kickoff.

₹8L–12L one-time Hardware not required — runs on Reyatech lab hardware or yours.
Team size
Up to 10 users
Duration
4 weeks
Support
Slack / email · business hours
  • Discovery + use-case scoping with eval rubric agreed up front
  • Fine-tuning on a sample of your data (typically 1–3 use cases)
  • Deployed PoC on Reyatech-hosted lab hardware (or your hardware)
  • Evaluation report against the agreed rubric, with sample outputs
  • 4-week support window — Slack/email response, working hours IST
Start With a Pilot
Pilot pass-bar guarantee

If the eval scores agreed in the SOW are not met at the end of the pilot, you pay zero. The pass bar is signed by both sides before kickoff — not at our discretion.

100% of pilot fee credited toward the Production tier if you proceed within 60 days of pilot close.
Recommended for 10–50 users
Production · 8 weeks

Production AI on your hardware, with everything you need to operate it.

Full LoRA fine-tune, on-prem deployment, audit-grade trail, integrations with your IdP / storage / SIEM.

₹15L–25L one-time + hardware (optional bundle) · see §5 below
Team size
10–50 users
Delivery
8 weeks from SOW signing
Post-launch
90-day stabilisation support
  • Full LoRA fine-tune on your data, across the agreed use cases
  • Deployment on your hardware (or our bundle); RAG pipeline; Keycloak SSO; Langfuse audit pipeline
  • Integration with up to 3 of your internal systems (storage, IdP, SIEM)
  • 8-week delivery · 90-day stabilisation support post go-live
  • IQ / OQ / PQ artefacts (Pharma) or §16/17 SOW exhibits (NBFC)
Book Discovery Call
Cure period: 30 days post-stabilisation for any unmet SOW commitment. Unresolved breach grants full refund of unrecognised milestones — see §9.
Enterprise · custom

Multi-department, multi-site, SLA-backed.

Production scope plus multiple use-case packs, custom integrations, and a dedicated engineering contact.

₹25L–40L+ one-time + hardware · scoped per engagement
Team size
50–500 users
Delivery
10–14 weeks · phased
SLA
Sev-1 1h · Sev-2 4h · 24×5
  • Everything in Production, scaled across departments
  • Multi-department use-case packs (e.g., QA + Regulatory + Manufacturing)
  • Custom integrations beyond the 3 included in Production
  • SSO + advanced RBAC, attestation-grade audit dossier
  • Dedicated engineering contact, SLA-backed response times
Talk to Us
Phased SOW: Enterprise engagements are split into named milestones. Each milestone is independently invoiceable and independently cancellable.
§3 · ROI calculator

Run it with your numbers.

Adjust team size, usage, horizon, and hardware — see TCO, savings, and breakeven update live. INR pre-GST. Assumptions: 22 working days/month, 70/30 input/output token split, 84 ₹/USD, GST 18%.

Cloud TCO —
OwnAI TCO —
Savings — —
Breakeven —
5-year TCO comparison: cloud baseline vs OwnAI on your hardware. Updates live from inputs above.
Component Yr 1 Yr 2 Yr 3 Yr 4 Yr 5 Total
Book Discovery Call Email this scenario as PDF

Estimates only — your costs will differ based on actual usage patterns, integration complexity, and electricity/rack space already amortised. Excludes one-time customer integration (10–25h depending on data volume). For a precise number under NDA, book a discovery call.

§4 · Why the range

Four variables drive where you land in the band.

Every quote is itemised. You see exactly what you're paying for, with the underlying assumptions documented in the SOW exhibits.

Data volume
More documents → more LoRA epochs → more cloud-GPU hours during training.
Use-case count
Each new task = its own eval rubric and adapter slice. 1 task ≠ 1/6 the work of 6 tasks — the floor is non-zero.
Integration depth
Pulling from SAP / Oracle / Tally costs more than pulling from a shared drive. We quote each integration discretely.
Concurrency target
More concurrent users at sub-200ms p50 latency can require a bigger hardware tier — quoted separately.
§5 · Hardware add-on

Hardware is yours. We configure and ship.

Sized to concurrency, not headcount. Standard manufacturer warranty applies. Or procure independently — we provide the BOM and acceptance checklist.

Tier 1 · 1–5 users
Mac Mini M4 Pro 64 GB
apple-silicon · inference only
₹2.4Lpre-GST
  • Model fitQwen 3 32B · Q4
  • Sustained14–18 tok/s
  • Power35 W typ / 80 W peak
  • Electricity≈ ₹3.5K / year
Tier 2 · 5–15 users
RTX 4090 Workstation 64 GB
single-gpu · workstation form-factor
₹4.2Lpre-GST
  • Model fitQwen 3 32B · Q4–Q8
  • Sustained38–55 tok/s
  • Power320 W typ / 600 W peak
  • Electricity≈ ₹26K / year
Tier 3 · 15–50 users
L40S 48 GB Server (1U)
rack-mount · datacentre-ready
₹8Lpre-GST
  • Model fitQwen 3 32B · FP16
  • Sustained60–85 tok/s
  • Power350 W typ / 700 W peak
  • Electricity≈ ₹30K / year
Hardware is yours. We configure and ship — sustained tok/s benchmarked on Qwen 3 32B @ 8K context. You may also procure independently — we will provide the BOM and acceptance checklist. Standard manufacturer warranty applies; AMC adds a spare-parts plan on top.
§6 · AMC

Annual maintenance — optional, recommended.

Think of it as a retainer for an AI team without the hiring overhead. Starts after the 90-day stabilisation window; cancellable on 90 days' notice.

Annual maintenance
18–22%
of setup fee · per year

The 4-point band reflects how aggressive your SLA is, how many refreshes you want, and whether your quarterly review is on-site or remote.

"A retainer for an AI team — without the hiring overhead."

Response-time SLAtier-defining
Sev-1 1-hour response (top of band) vs Sev-1 4-hour response (bottom of band). Sev-2 / Sev-3 thresholds scale accordingly.
Model refreshes per yeardefault 4
Default 4 quarterly refreshes; can negotiate to 2 (lower) or 6+ (higher).
CVE patching cadencedefault monthly
Default monthly CVE sweep + patch for all OSS deps. Weekly cadence available at top of band.
Quarterly reviewremote / on-site
Remote (bottom) vs on-site at your premises (top). Includes usage analytics, drift detection, refresh planning.
Out of scope · priced separately
  • New use cases beyond the original scope
  • New verticals
  • Hardware replacement beyond manufacturer warranty
§7 · What's not included

Said out loud, so it isn't a surprise.

Items that look adjacent but aren't part of the quoted fee.

Per-seat licence for any commercial LLM. We don't use them. There is no per-seat cost to pass through.
Cloud GPU costs for fine-tuning runs. We absorb these — they are part of the project fee.
Customer-side change management or training programs. Separate SOW, priced by scope and trainer count.
Validation services beyond the artefacts listed in each tier. e.g. independent third-party CSV. Quoted as a separate SOW with the validation partner.
GST (18%). Applied on top of all quoted prices. Every invoice itemises CGST/SGST or IGST per applicable rule.
§8 · Payment terms

Three milestones. No surprise invoices.

Each instalment ties to a SOW-defined event. AMC is billed annually in advance — start date = go-live + 90 days.

Step 01 · Kickoff
50%
On SOW signing
Triggers fine-tuning and deployment work. Net 15 from invoice date.
Step 02 · Deployed
25%
On production deployment
Defined in SOW — system reachable on your LAN, IQ artefacts shipped.
Step 03 · Live
25%
On go-live sign-off
Defined in SOW — PQ certificate, runbook handover, AMC start date locked.
Step 04 · AMC
+AMC
Annually, in advance
From go-live + 90 days. Cancellable on 90 days' notice — no auto-renewal traps.
§9 · Refund & exit

If we don't deliver, you don't pay.

The exit terms that get auditors and procurement teams comfortable. All three apply by default — they're in the standard SOW.

Pilot tier
Pass-bar guarantee

If the eval scores agreed in the SOW are not met at the end of the 4-week pilot, you pay zero. The pass bar is signed before kickoff — not at our discretion.

Production tier
30-day cure period

Post-stabilisation, a 30-day cure period is granted for any unmet SOW commitment. Unresolved breach grants the customer full refund of unrecognised milestones.

AMC
90-day cancellation

Cancellable on 90 days' notice at the customer's option. No auto-renewal traps. The system keeps running — only the support contract ends.

§10 · FAQ

The seven questions procurement asks.

What's the payment schedule?
50% on SOW signing → 25% on production deployment → 25% on go-live sign-off. AMC is billed annually in advance from go-live + 90 days. Each milestone is defined in the SOW and tied to a named deliverable; see §8.
Is hardware included?
Optional bundle — see §5. You may also procure independently; we'll provide the BOM and acceptance checklist. Standard manufacturer warranty applies. AMC adds a spare-parts plan on top of the warranty.
What if the pilot doesn't work?
You pay zero, per the pass-bar criteria signed at kickoff — not at our discretion. The eval rubric is countersigned by both sides before any work begins.
Can we start with one department?
Yes. Most customers start with 10–25 users in one function — typically QA in pharma or credit/compliance in NBFC. The Pilot tier is designed for exactly this shape of engagement.
Is GST included?
No. All prices are pre-GST. 18% GST applies on every invoice — itemised as CGST/SGST or IGST depending on whether the customer's billing state matches our registered state (Karnataka).
What happens after the AMC expires?
The system keeps running. You lose access to quarterly model refreshes and priority support — nothing else. The base models are open source; the adapter is yours; the source code is escrowed.
Can the price be lower for a smaller scope?
Yes — a single-use-case pilot can land at the bottom of the Pilot range. We will also tell you when scope is too small to be worth the engagement — it's a worse outcome for both sides if the project can't move the needle.
Not sure which tier?

Let's scope it together.

A 30-minute call to talk through your use cases, your team size, and the regulatory frame you operate in. We'll send a pre-quote within 48 hours.

Book a 30-Minute Discovery Call Email the founder
rishi@reyatech.com +91‑7486461783
All prices INR · pre-GST Pilot pass-bar guarantee No auto-renewal