elan1 Run Contracts
Ten fusion.*/v1 contracts, each with two licence-clean implementations.
Contracts in detailInfrastructure Intelligence Platform
elan1 Run is the infrastructure intelligence platform for the open-source AI stack. It puts ten versioned fusion.*/v1 contracts between your applications and components such as model servers and vector stores, proves them with an offline conformance suite, swaps components through human-gated substitution and runs Day-2 operation, priced by node and GPU, never by token.
01What it does
elan1 Run Runtime Controls, with elan1 Govern built in.
The named profile for regulated teams.
elan1 Run puts each part of the open-source AI stack behind a versioned contract. Each contract has two licence-clean implementations and a conformance suite. A nine-step substitution process changes components, and Day-2 operation keeps the stack running.
Who it’s for
02How it works
Your applications call the contract. The component behind it can change, under conformance and with people at the switch.
03Key capabilities
Each module carries its status. Nothing here is shipped.
Ten fusion.*/v1 contracts: chat, gateway, inference, retrieval, vectors, trace, tools, run, memory and guard. Examples behind them include LiteLLM, vLLM, pgvector, Langfuse and Temporal.
Suite K1–K7, blocking-green, runs offline, including on your auditor's machine.
Nine steps, with human gates H1 and H2. It is never autonomous.
A budget hard-stop before the first token, plus attribution by tenant, team and use case. A guard covers PII, provenance and injection. Tool authorisation has an autonomy ceiling. A licence register pins each component at its exact SHA, and an isolation attack runs nightly.
Six runtime units are counted: tokens in, tokens out, GPU-seconds, bytes, pages and tool calls.
Direct, through a systems integrator, or as Operator OEM, where the operator's brand is the only brand shown. elan1 Run Cloud (up to 25 tenants), customer VPC and on-prem. Air-gapped is on What's next.
04Modules
Every module is included in every subscription. Tokens and the six runtime units are counted and shown, never priced.
Ten fusion.*/v1 contracts, each with two licence-clean implementations.
Contracts in detailSuite K1–K7, blocking-green, runs offline, including on your auditor's machine.
Conformance in detailNine steps, with human gates H1 and H2. It is never autonomous.
Substitution in detailDay-2 operation. Six runtime units are counted: tokens in, tokens out, GPU-seconds, bytes, pages and tool calls.
Operate in detailelan1 Run Runtime Controls, with elan1 Govern built in. A budget hard-stop before the first token, attribution, a guard and tool authorisation.
Runtime Controls in detailelan1 Run governs the call. elan1 Govern governs the write.
Run pricing by module05Two kinds of governance
Every control in Run’s contract layer is a promise about a call: who made it, what it cost, which component served it and what data it touched. None of them decides whether a business write may happen. That belongs to elan1 Govern alone. Run asks Govern for a gate check at substitution step 7.3, and honours the answer. A regulated team can buy one thing and still keep the two jobs apart.
Named profile
elan1 Run Runtime Controls, with elan1 Govern built in.
elan1 Run can prove a call happened. Only elan1 Govern can prove it was allowed.
A guard verdict is an input, never the allow.
The approval wait is held on the run contract, and expiry means refusal.
Trace spans bind to the proof record.
Substitution step 7.3 calls a Govern gate check.
Erasure is itself a governed write.
Restore drills are verified with elan1-verify.
Choose a seam number to pin both ends of it.
06Inside the product
Contracts and their ranked bindings, the nightly conformance matrix, a substitution at its human gates, a trace, the attribution explorer and the licence register. Choose a screen; each lists what to look for.
01 / 14 · Home

02 / 14 · Contracts
03 / 14 · Contracts
04 / 14 · Conformance
05 / 14 · Conformance
06 / 14 · Substitutions
07 / 14 · Substitutions
08 / 14 · Substitutions
09 / 14 · Traces
10 / 14 · Traces
11 / 14 · Meter
12 / 14 · Trust
13 / 14 · Trust
14 / 14 · Operations
Illustrative specimen. Not customer data. Screens come from the elan1 Run console running on fictional sandbox data.
07Deployment
Three ways to buy, three places to run it. On an OEM console, the operator's brand is the only brand shown.
Your platform team runs the stack with us: a paid 90-day pilot, then a subscription by node-year.
An SI delivers and operates the stack for its clients, licensed for ten deployments a year.
A GPU operator offers the managed AI stack under its own brand. The operator’s brand is the only brand shown.
| Where it runs | What it means | Status |
|---|---|---|
| elan1 Run Cloud | Hosted by elan1, for up to 25 tenants. | Available |
| Customer VPC | In your own cloud account. | Available |
| On-prem | In your own data centre, on your own GPUs. | Available |
| Air-gapped | No outside network at all. | What’s next |
The operator’s view


Illustrative specimen. Not customer data. Fictional sandbox data from the console.
08Assurance level and mode
Available now.
Run attests calls, components, licences and conformance. It never attests business actions. Its own tooling writes to Govern in Mode B: evaluated and recorded, failing closed if elan1 is unreachable.
When Run's tooling makes a write, it goes to elan1 Govern in Mode B. If elan1 is unreachable, it fails closed.
Business actions belong to elan1 Govern. See elan1 Govern
09Meter and pricing
The unit
One node-year
Per node and per GPU. Runtime units are counted and shown, never priced.
Priced by capacity, never by tokens. The six runtime units are counted and shown, never priced. A call refused at the edge returns a 409 before the first token. It costs 40% of one governed action: ₹0.64 at the L1 rate.
Every subscription includes all five modules: Contracts, Conformance, Substitution, Operate and Runtime Controls. The Runtime Controls governance profile, with elan1 Govern built in, adds three nodes under management (below).
| 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 |
| What | Price · excl. GST | Includes |
|---|---|---|
| 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 |
| 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 |
| What | Price · excl. GST | Includes |
|---|---|---|
| Priority support | 25% of the subscription | |
| 24×7 named-engineer support | 40% of the subscription | |
| elan1 Run Runtime Controls, with elan1 Govern built in | ₹5,40,000 per year | The Runtime Controls module is included in every subscription; the governance profile adds three nodes under management (contracts C2, C6, C7, C8, C10) |
| Paid 90-day pilot | ₹4,50,000 | |
| Onboarding — Standard | ₹2,10,000 per new account | |
| Onboarding — Sovereign | ₹8,00,000 per new account |
All prices are exclusive of GST. Prices are current list prices; GST is added at the applicable rate on the invoice.
10Where it is weak
Plain limits, so you can plan around them.
11Status
Every elan1 Run subscription includes all five modules.
Available now. Future improvements are listed, in order and without dates, on What’s next.
Announced 10 Oct 2026: elan1 Govern, Work, Decide and Run are available.
12FAQ
The components that serve AI applications: model gateways and inference servers, retrieval and vector stores, tracing, tool calling, workflow runs, memory and guards. Examples behind the ten fusion.*/v1 contracts include LiteLLM, vLLM, pgvector, Langfuse and Temporal. elan1 Run operates them under contracts rather than replacing them.
Yes. Your application calls a fusion.*/v1 contract, not the component behind it, so elan1 Run Substitution can swap a model or vector store through nine steps with two human gates, H1 and H2. Our target is inside 72 hours with no change to consumer code; it is a target, not a published measurement.
Per node-year, never per token. Starter is ₹1,80,000 a year with two CPU nodes included; Standard is ₹6,00,000 and Sovereign ₹24,00,000 a year, plus ₹1,80,000 a year per CPU node and per GPU under management. Tokens are counted and shown, never priced, and all prices exclude GST.
Yes. elan1 Run deploys as elan1 Run Cloud, in your own VPC or on-premises, and the Sovereign subscription is ₹24,00,000 a year plus capacity under management. Air-gapped deployment is not available today; it is on What's next.
It governs the runtime: calls, cost, components, licences and substitution. Approving business actions is elan1 Govern's job, and Govern comes built in with Runtime Controls.
No. On an OEM console, the operator's brand is the only brand shown.
No. Run never reads tenant records.
Next steps
Related reading
One write path, four layers
Work and Decide propose. Govern and Run assure. Only Govern allows.
// elan1 RUN
Start with a paid 90-day pilot: your stack behind the contracts, conformance running nightly, every call counted.