Key facts
| API compatibility | OpenAI-compatible /v1/chat/completions and /v1/embeddings |
| Models | 30+ models across free, paid and embedding tiers behind one endpoint |
| Agents and tools | Function calling is live; assistants and responses endpoints are coming soon |
| Knowledge | RAG and embeddings for private knowledge assistants |
| Deployment | Plugsky cloud, VPC, on-prem, air-gapped and bring-your-own-cloud |
| Regions | Region pinning including GCC, EU, US and APAC options |
| Free plan | 2 free AI models (plugsky-micro, plugsky-lite), 2 API keys, no card |
| Trial | 14-day full-access trial with models up to plugsky-plus, no card |
TL;DR
- One OpenAI-compatible API for 30+ models, embeddings and tools.
- Start free with plugsky-micro and plugsky-lite, then scale on flat plans.
- Agents run today on function calling; Assistants-style endpoints are coming soon.
- Region pinning and sovereign deployments cover regulated workloads.
- Migrate by changing the base URL and model name — rollback is the same edit.
How it works, step by step
- Create an account and generate a scoped API key on the free plan.
- Point your OpenAI SDK at https://api.plugsky.com/v1 and send a first chat completion.
- Pick models per workload from the catalogue rather than one model everywhere.
- Add embeddings and RAG when answers need your own documents.
- Add function calling when the model must act, and structured output when code must parse.
- Choose a region and deployment model that satisfies your data rules.
- Scale to a paid plan or an enterprise agreement when limits or controls require it.
Original data
Try it yourself
Open the OpenAI-compatible API tester →
The API layer
Plugsky speaks the OpenAI /v1 namespace, so the migration is one line:
from openai import OpenAI
client = OpenAI(api_key="sk-live-…", base_url="https://api.plugsky.com/v1")
resp = client.chat.completions.create(
model="plugsky-pro",
messages=[{"role": "user", "content": "Say hello in 5 languages"}],
)Streaming, function calling, JSON mode and embeddings all use the same shapes you already know. Python, Node, Go, Java, Rust and raw cURL examples are documented, and framework integrations cover LangChain, LlamaIndex, Vercel AI SDK, Haystack, Semantic Kernel and AutoGen.
The model catalogue and platform products
One subscription reaches 30+ models: small and fast free tiers, mid-range production workhorses and frontier reasoning models, plus embedding models for retrieval. Platform products around the API include agents and tools on function calling, knowledge and RAG for private assistants, model fusion for multi-model workflows, usage analytics, webhooks and scoped keys. The assistant-family endpoints are coming soon, so agent state lives in your own database today — which also keeps it portable.
Sovereign and regulated deployment
- Cloud: the fastest start, with region pinning to keep data in the selected geography.
- VPC: Plugsky runs inside your AWS, GCP, Azure or OCI network with your KMS and peering.
- On-prem and air-gapped: Helm or physical-media deployment for environments with no internet path.
- Bring-your-own-cloud: control plane managed, inference in your own cloud account.
- White-label: your brand and domain on top of the platform for SaaS resellers.
Encryption in transit and at rest, customer-managed keys, SSO/SCIM, RBAC and audit export support the governance side; the published SLA lives at /legal/sla and the terms at /legal/terms.
Where Plugsky does not replace what you have
Honest boundaries: image generation, audio, moderation, files, batch, fine-tuning and Assistants-style endpoints are coming soon rather than available today, so requirements that depend on them need a different plan or a wait. Completed SOC 2 and ISO audits are in progress rather than finished — verify current status and compensating controls during diligence. Hyperscalers offer broader native governance suites. Plugsky's advantage is the combination of an OpenAI-compatible API, flat-rate self-serve pricing, 30+ models and sovereign deployment in one platform; if you need an endpoint that is not live yet, check /docs before committing.
Honest comparison
| Dimension | Plugsky | Single-model provider | Self-hosted stack |
|---|---|---|---|
| API surface | OpenAI-compatible /v1 with extensions | Vendor-specific | You define it |
| Model choice | 30+ models, one key and invoice | One model family | Whatever you host |
| Deployment | Cloud, VPC, on-prem, air-gapped, white-label | Provider cloud | Your infrastructure |
| Agents today | Function calling on chat completions | Varies | You build everything |
| Cost model | Flat monthly plans, fair-use usage | Per-token | GPU plus operations |
| Ops burden | Managed | Managed | High |
Frequently asked questions
What is Plugsky in one sentence?
An OpenAI-compatible AI platform that provides 30+ models, embeddings, agents and RAG through one API, with flat-rate plans and sovereign deployment options.
Is Plugsky OpenAI-compatible?
Yes. The chat completions and embeddings endpoints follow the OpenAI /v1 contract, so existing SDK code works after changing the base URL and model name.
What does the free plan include?
Two free AI models, plugsky-micro and plugsky-lite, with two API keys and no credit card. A 14-day full-access trial is also available. See the live pricing page for current details.
Which models can I use?
30+ models across free, paid and embedding tiers, all behind one endpoint and one API key. Browse the live catalogue at /models for context limits and capabilities.
Does Plugsky have agents?
Function calling is live on chat completions, which is enough to build tool-using agents. Assistants-style and responses endpoints are documented as coming soon.
Can I keep data in my own country?
Yes. Region pinning keeps data in the selected geography, and VPC, on-prem, air-gapped and bring-your-own-cloud options go further for regulated requirements.
How does Plugsky handle data privacy?
Scoped keys, encryption, customer-managed key options, audit logs and DPA terms underpin privacy; training-use and residency commitments are defined in the platform terms. Confirm wording at /legal/terms.
What is not available yet?
Image generation, audio, moderation, files, batch, fine-tuning and Assistants-style endpoints are coming soon, and completed SOC 2/ISO certifications are in progress. Check /docs and the status page before planning around them.