~/f4n6 $ adversetrace --services --status open
STATUS: accepting engagements JURISDICTION: UK-resident inference STACK: open-weights · vLLM · zero egress
■ TLP:CLEAR — SOVEREIGN INFERENCE

services

// honest scope

What this is — and what it isn't

Sovereign appliances solve a specific problem extremely well. They are not a drop-in replacement for a hyperscale API — and knowing the difference up front is the point.

▸ BUILT FOR

  • Regulated workloads where data cannot leave your jurisdiction or network
  • Internal RAG, document intelligence and case analysis over sensitive corpora
  • Predictable, fixed monthly cost instead of unbounded per-token spend
  • Firms that need the model never to be deprecated, throttled or revoked beneath them
  • Human-in-the-loop pipelines with a complete, exportable audit trail

✕ NOT FOR

  • High-QPS public-facing chat at hyperscale concurrency
  • Always needing the single newest closed-weight frontier model
  • Spiky, unpredictable seasonal load better served by elastic cloud
  • Teams with no appetite for a managed appliance or support relationship
// engagement models

Three ways to buy

Indicative pricing. Every engagement is scoped against your current token spend, the cloud-reserved equivalent, and your regulatory obligations.

Project

Build & handover

from £15k one-time
We scope, build and harden a sovereign inference appliance on your premises — then hand you the keys and the documentation. You own the hardware and the stack outright.
  • Hardware specced & supplied at cost-plus
  • vLLM serving + model selection + RAG
  • Security hardening & documentation
  • DORA / NIS2 control-mapping pack
  • Optional knowledge transfer
scope a build
★ most chosen

Managed appliance

£2k–£6k / month
A dedicated, single-tenant appliance — at your site or in our UK rack — that we run for you end to end. You get the capability; we carry the operational burden and the SLA.
  • Single-tenant — never shared
  • Model swaps & version management
  • Patching, monitoring, capacity tuning
  • Tiered support SLA
  • 24–36 month term, hardware amortised
request a quote
Subscription

Hosted endpoint

£1.5k–£4k / month
Dedicated, UK-resident capacity on our cluster behind an OpenAI-compatible API. The fastest route to a sovereign endpoint — no hardware to buy, no procurement lead time.
  • Single-tenant reserved capacity
  • OpenAI-compatible API
  • Priced at or under your token bill
  • UK data residency guaranteed
  • No procurement lead time
start a trial
// cost model

Fixed beats unbounded

The pitch isn't “cheaper than the API on day one” — it's a known number you control, with sovereignty the cloud can't sell you. Illustrative; we model your real workload during scoping.

Dimension
Sovereign appliance
Per-token cloud API
Monthly cost
Fixed & predictable
Scales with usage
Data residency
Your network / UK
Provider-controlled
Model lifecycle
You decide
Can be deprecated
Audit trail
Full, exportable
Limited
Best at
Sustained, sensitive load
Spiky, public load
// regulatory framing

Built by people who live in the regulations

A sovereign DFIR & OSINT practice serving EMEA financial services. Sovereignty isn't a feature bolted on — it's the operating assumption.

DORA Art. 17 / 19 / 28–30
NIS2 Art. 21(2)(d) / 23
CLOUD Act exposure removed
Human-in-the-loop review before delivery
// scope it

Scope your sovereign appliance

A 30-minute call to map your workload, your obligations and the right engagement model. No obligation, no token meter running.

hello@f4n6.co.uk →