elan1 RUN · OPERATE
Day 2 is what you are paying for.
elan1 Run Operate is Day-2 operation for the open-source AI stack: running, patching and observing it after go-live. It counts six runtime units, tokens in and out, GPU-seconds, bytes, pages and tool calls, and shows them on the bill without pricing them. You pay per CPU node and per GPU under management.
Available now, in every elan1 Run subscription.
01What it does
Five runbooks, each ending in a command that checks it.
- Metering
- Six runtime units: tokens in, tokens out, GPU-seconds, bytes stored, pages ingested and tool calls. Each is tagged once, at the edge, with tenant, team and use case. All are counted and shown; none is priced.
- Change
- Scheduled upgrades ride a train with ten working days' notice. Forced changes go through Substitution.
- Incidents
- Rehearsed runbooks, including India's CERT-In six-hour reporting clock.
- Restore
- Restore drills are verified with elan1-verify against Govern’s proof records: a pass count, not a row count.
- Exit
- Your state leaves in published custody formats. An exit that needs our cooperation is not an exit.
Metering, for one tenant
Counted and shown. Never priced.
Each unit is tagged once, at the edge, so the same rows answer for the tenant, the team and the use case. Choose how to attribute them.
Kolam Components* · one month
tagged once, at the edgetenant · team · use casetenant=kolam
| Unit | Counted | Price |
|---|---|---|
| Tokens in | 18,402,116 | Not priced |
| Tokens out | 4,118,930 | Not priced |
| GPU-seconds | 212,640 | Not priced |
| Bytes stored | 1,842 GB | Not priced |
| Pages ingested | 36,210 | Not priced |
| Tool calls | 9,874 | Not priced |
Billed: capacity, by node-year.Six units counted and shown. None is priced.
02Who it's for
Teams who run the stack after go-live.
- GPU operators selling a managed AI stack under their own brand.
- SIs who hand a stack over and still need it run.
- Regulated platform teams who must know what runs, what it cost and who used it.
03How it works
One artefact, five places to run it.
One artefact runs in every mode, bought direct or through an SI. India comes first, in AWS ap-south-1 (Mumbai). New regions, the Gulf and the EU among them, open only on a signed contract, never on a forecast.
| Mode | What it is |
|---|---|
| elan1 Run Cloud | Starter and pilots, up to 25 tenants |
| Customer VPC | In your cloud account |
| On-prem | In your data centre |
| Air-gapped | On What's next |
| Operator OEM | On the operator's capacity; the operator's brand is the only brand shown |
04In the console
Day 2, as the console shows it.
Every metered call attributed by team and use case, the objectives the on-call operator works to, and where each copy of a tenant's data lives.
Illustrative specimen. Not customer data. Fictional sandbox data from the console design.
05What it connects to
Operate keeps Govern's service running. It never decides for it.
Operate keeps elan1 Govern’s service running on the same runtime but never decides for it. Writes go through elan1 Govern. Erasure is itself a governed write, and a tenant stop during an incident is made through Govern at rung 0b.
Writes go through elan1 Govern.
06What it costs
Priced by capacity, never by tokens.
The unit
One node-year
Per node and per GPU. Runtime units are counted and shown, never priced.
Operate is included in every elan1 Run subscription. Priced by capacity, never by tokens: every CPU node and every GPU under management is a line on the bill.
| What | Price · excl. GST | Includes |
|---|---|---|
| Starter | ₹1,80,000 per year | 2 CPU nodes included |
| Standard | ₹6,00,000 per year | Plus nodes and GPUs under management |
| Sovereign | ₹24,00,000 per year | Plus nodes and GPUs under management |
| CPU node under management | ₹1,80,000 per node per year | |
| GPU under management | ₹1,80,000 per GPU per year | |
| Tokens and runtime units | ₹0 — counted, never priced | Tokens in and out, GPU-seconds, bytes, pages, tool calls |
| Paid 90-day pilot | ₹4,50,000 | |
| A call refused at the edge | 40% of one governed action (₹0.64 at the L1 rate) | Returns a 409 before the first token, so no token is sent |
All prices are exclusive of GST. Prices are current list prices; GST is added at the applicable rate on the invoice.
- Support is its own line: priority support, or 24×7 named-engineer support, each a share of the subscription.
- Systems-integrator and operator OEM terms are on the pricing page.
All prices are exclusive of GST. GST is added at the applicable rate on the invoice.
07Status
Available now.
Air-gapped deployment is on What's next.
- Runtimeelan1 Run is the real runtime for elan1’s own layers.Status: Available
- PilotA paid 90-day pilot, on your nodes or ours.Status: Available
- OEMOperators can run elan1 Run under an OEM licence. Partner terms on request.Status: Available
08What it does not do
What Operate does not do.
- It does not read your business records.
- It does not own GPUs, resell inference or bill per token.
- It publishes no recovery point or recovery time objective.
- It has no SOC 2 report or ISO 27001 registration today.
09FAQ
Six direct answers.
Running the open-source AI stack after go-live: metering, upgrades, incidents, restores and exit, each handled as a runbook. elan1 Run Operate also counts six runtime units per call and shows them on the bill.
Because elan1 Run is priced by capacity, never by tokens. Tokens in and out, GPU-seconds, bytes, pages and tool calls are counted and shown so you can see what every call cost, but the bill is per node and per GPU.
₹1,80,000 a year per CPU node and ₹1,80,000 a year per GPU under management, on top of a Starter, Standard or Sovereign subscription. All prices are exclusive of GST.
No. On an OEM console, the operator's brand is the only brand shown.
Yes, in your VPC or on-prem. Air-gapped is on What's next.
It leaves with you, in published formats you can read without us.
10Next steps
Where to go from here.
Next steps
One write path, four layers
Work and Decide propose. Govern and Run assure. Only Govern allows.
// elan1 RUN · OPERATE
Day 2 is what you are paying for.
Start with a paid 90-day pilot: the stack run for you, every unit counted and shown, every runbook rehearsed.