Key facts
| Data residency | Region-locked data planes; deployment model selected per workspace — see /data-residency |
| API compatibility | OpenAI-compatible /v1/chat/completions; change the base URL only |
| Models | 30+ models behind one API, from free tiers to frontier reasoning |
| Deployment | Plugsky cloud, your VPC, on-prem and air-gapped options |
| Pricing model | Flat monthly self-serve plans with unlimited fair-use usage; no per-token billing on those plans |
| Free tier | plugsky-micro and plugsky-lite on the free plan, no card required |
| Identity and audit | SSO/SCIM, role-based access, audit log export and BYOK via KMS or HSM |
| Local presence | No office or data center in Edinburgh claimed; residency runs through region plans and private deployment |
TL;DR
- Keep your OpenAI SDK code — change the base URL and model name.
- Pick a region-locked data plane that matches your jurisdiction; details at /data-residency.
- 30+ models behind one API, from free tiers to frontier reasoning.
- Deploy in Plugsky cloud, your VPC, on-prem or air-gapped as policy requires.
- Flat monthly self-serve pricing; start on the free plan and move up when ready.
How it works, step by step
- Map the data stores, logs and subprocessors your Edinburgh workload touches.
- Agree the residency boundary: jurisdiction, region and deployment model.
- Create a Plugsky account and generate an API key on the free plan (no card).
- Point your OpenAI-compatible client at the Plugsky base URL and map model names.
- Run your evals on real Edinburgh prompts, including latency and output-quality checks.
- Confirm audit log export and deletion evidence with security, then cut over production traffic.
Try it yourself
Work through the residency checklist →
What Edinburgh enterprises should expect from an AI cloud
Edinburgh is Scotland's capital and a centre for asset management, insurance, public administration and university research, with buyers who routinely require documented data handling. Teams there increasingly ask where inference data lives, which subprocessors can reach it, and whether failover can move it across a border. Those are architecture and contract questions, not marketing claims.
- Financial services: asset managers and insurers expect model risk controls and audit evidence.
- Public sector: procurement frameworks require clear processing locations and exit terms.
- Research and education: grant-funded projects need reproducible, boundary-respecting infrastructure.
What to validate before you commit
A region name is not an answer on its own. Test the four layers underneath it:
- Where data rests: the store-by-store map, not a single headline region.
- How it moves: private connectivity, encryption, and log destinations across regions.
- Who can reach it: named subprocessors and upstream model providers, with change notice.
- What happens at exit: portable export, verifiable deletion, and audit records you can retain.
Plugsky documents region-locked data planes and deployment options at /data-residency. Confirm the current region list there before you commit, and get the chosen boundary written into your DPA.
Where Plugsky helps in Edinburgh, and its limits
Plugsky is built to drop into an existing OpenAI-compatible codebase: swap the base URL, map model names, keep your evaluation suite. From there you get 30+ models on one API, flat monthly self-serve pricing, and deployment from shared cloud through VPC and on-prem to air-gapped.
Be clear about the gaps — audio, images, fine-tuning and batch endpoints are roadmap items, and a self-serve plan does not create a physical presence in Edinburgh. A dedicated in-country footprint is a bespoke engagement with its own residency and operations plan, not a checkbox on a signup form.
Honest comparison
| Capability | Plugsky | Typical global AI API | Building in-house |
|---|---|---|---|
| API compatibility | OpenAI-compatible; one-line base URL change | Usually OpenAI-compatible | Full rewrite and retest |
| Data residency | Region-locked planes plus VPC, on-prem and air-gapped | Fixed set of vendor regions | You control and you operate it |
| Model choice | 30+ models behind one API and one key | Vendor catalogue only | You host and upgrade each model |
| Pricing | Flat monthly self-serve plans, fair-use usage | Usually per-token | GPU, power, staff and ops |
| Time to first call | Minutes on the free plan, no card | Minutes, card usually required | Weeks to months of platform work |
Frequently asked questions
Does Plugsky have an office or data center in Edinburgh?
No local office or data center in Edinburgh is claimed. Residency is delivered through region-locked data planes and deployment options — Plugsky cloud, your VPC, on-prem or air-gapped — selected per workspace. The current region list is documented in the data residency docs.
Can we keep United Kingdom workloads inside our chosen region?
You can select a region-locked data plane that matches your jurisdiction and add private deployment when policy requires it. Verify the current region list at /data-residency and have the choice recorded in your DPA before go-live.
Can we keep using the OpenAI SDK?
Yes. The chat completions endpoint is OpenAI-compatible, so you change the base URL and model name and keep your existing SDK code, prompts and tests.
How does pricing work?
Self-serve plans are flat monthly with unlimited fair-use usage and no per-token billing on those plans; see the live pricing page for current tiers. Enterprise deployments are quoted separately.
Is there a free plan?
Yes — plugsky-micro and plugsky-lite are available on the free plan with no credit card, and a 14-day full-access trial is available for evaluating the wider catalogue.
Can we deploy inside our own network?
Yes. Deployment options include your VPC, on-prem and air-gapped environments for teams with strict boundary or connectivity requirements.
How do we verify residency claims before signing?
Ask for a store-by-store residency map, the subprocessor list, failover behaviour and deletion evidence, then run a pilot with your own prompts and inspect the audit logs before committing.
Can we start small and scale later?
Yes — begin on the free plan or a self-serve tier, prove the workload, then move to enterprise deployment options when residency, scale or support requirements justify it.