Your model, on your data, in 72 hours.

The pilot, hour by hour (sample schedule)

Open-weight LLMs deployed inside your network, on GPUs you control, behind an OpenAI-compatible endpoint. Scroll to run the clock through a sample schedule.

Start the $4,900 pilot Talk to an engineer

T+00:00Clock startsSample schedule

Access and prerequisites

The 72 hours begin once we have access to your environment and the prerequisites are in place. Not before.

Access to your environmentneeded
Prerequisites in placeneeded
72-hour clock startsT+00:00

Deliverable: confirmed start. Fixed price $4,900.

T+08:00Model pulledSample schedule

One open-weight model, chosen for your workload

We pick from the models we deploy and load it onto your GPUs. Nothing is sent to a third-party API.

Sample logweights fetched ok
checksum verified ok
loaded on GPU 0 in your network

Deliverable: the model, running on your hardware.

T+20:00QuantizedSample schedule

Sized to fit the hardware you have

The model is quantized and tuned to run on the GPUs you control, so the pilot reflects what production would cost you.

Quantized to fit your GPUsdone
Batching and caching tuneddone
Throughput measured on your hardwarenext

Deliverable: a model sized to your GPUs.

T+36:00Endpoint upSample schedule

An OpenAI-compatible endpoint inside your network

Your applications talk to it the way they talk to OpenAI today.

client = OpenAI(
    base_url="https://llm.yourcompany.internal/v1",   # inside your network
    api_key="your-key",
)

Same SDK. Change the base URL.

Sample pingPOST /v1/chat/completions   
served from inside your network

Deliverable: a working OpenAI-compatible endpoint.

T+54:00EvaluationSample schedule

A reference evaluation on your data

The model is measured on your own material, so the decision rests on your results and not on a public leaderboard.

Quality on your own tasksmeasured
Response speedmeasured
Cost per requestmeasured

Deliverable: reference evaluation on your data.

T+72:00ReadoutSample schedule

Readout and production scoping

You leave with a measured result on your data and a scoped plan for production. The pilot costs $4,900, fixed.

Measured result on your datahanded over
Scoped plan for productionhanded over

Next rung: production deployment from $25,000

What you get at the end: the readout

The last hours produce a written result on your data and a check of the network boundary. The traces here are demo telemetry that show the format, not your numbers.

Reference evaluation: Quality on your own tasks, response speed and cost per request, measured on your data.
Egress scan: Every external endpoint the deployment contacts, listed. In the sample on the right the count is 0.
Production scoping: A scoped plan, from $25,000. We size the hardware and source it at cost, with no markup.

CH1 requestsCH2 GPU loadCH3 egress

DEMO TELEMETRY

Egress scanDEMO

external endpoints0

Scan complete. 4 internal nodes.

What we need before the clock starts

The 72 hours begin once access and prerequisites are in place, so the schedule above is never spent waiting on us or on you.

Access to your environmentWe work inside your network, so we need a way in to the GPUs and the systems around them.
GPUs you controlThe model runs on hardware you own or rent, behind your own network boundary.
Your data to test onThe reference evaluation is run on your material, so we need a representative sample to measure.

Talk to an engineer.

Tell us the model, the data and the GPUs you have. An engineer from Limassol or Auckland scopes the 72-hour pilot with you.

Start the $4,900 pilot Talk to an engineer

limassol.officeLocal time

Limassol HQ

Cyprus, 24/7

+357 25 123877

auckland.officeLocal time

Auckland

New Zealand, 24/7

+64 9 887 9088