Short answers to the questions buyers, developers and procurement teams ask most. Last reviewed 19 November 2026.
Product and models
What is Plugsky?
Plugsky is a sovereign AI cloud with an OpenAI-compatible API, 39 models on one endpoint, and in-region hosting. It also operates a GPU Share Network for owner-hosted capacity. You can start on the free tier without a card.
Is the Plugsky API OpenAI-compatible?
Yes. Set base_url to https://api.plugsky.com/v1 and use your Plugsky API key. Existing OpenAI SDK code generally works after those two changes; the docs list the supported endpoints and any model-specific differences.
Which models are available?
The catalogue has 39 models. The six Plugsky tiers are micro, lite and plus (30B total parameters, about 3B active) and pro, max and ultra (550B total, about 55B active), all with up to 262,144 tokens of context. See all models for per-model limits.
Do you support embeddings and RAG?
Yes. Plugsky Embed is a 2,048-dimension model with an 8,192-token input limit, and Plugsky Embed Multilingual covers cross-language retrieval. The embedding comparison covers dimensions, storage and cost.
Pricing and billing
How much does Plugsky cost?
Free is $0, Hobby is $8/month, Starter is $20/month, Builder is $60/month and Scale is $120/month. Annual billing is 20% less than 12 monthly payments. The full cap table is on pricing.
Is there a free tier?
Yes. Free requires no card and includes 500 requests/day (100 for new accounts), 250,000 tokens/day (50,000 for new accounts), 2 concurrent requests (1 for new accounts) and the Plugsky AI engine models micro and lite.
Which plan should I choose?
Hobby fits low-volume prototypes and small automations. Starter adds 2,000 requests/day, a 240 RPM limit and 6 concurrent requests. Builder and Scale are for higher volume with unlimited fair-use or unlimited daily terms. Estimate your bill with the LLM API cost calculator and your headroom with the rate-limit calculator.
What does “unlimited fair use” mean?
Builder and Scale do not have a daily request or token cap, but they still have RPM, concurrency and maximum input/output limits. Builder is 600 RPM and 10 concurrent requests; Scale is 1,200 RPM and 16 concurrent. Scale is described as unlimited usage. Check the plan table before committing.
Is there a student discount?
Yes. Verified students get 90% off inference credits for 12 months. Verification and eligibility are on the education discount page.
Is there a money-back guarantee?
Every paid plan includes a 7-day money-back guarantee on the first paid month. Contact support within 7 days of the first charge for a refund to the original payment method. Abuse, resale and usage beyond the fair-use limits are excluded.
Can I cancel at any time?
You can cancel from Dashboard → Billing. Access continues to the end of the paid period. Crypto annual prepay is a one-time payment with no auto-renewal. See the terms for the effective date and any exclusions.
Which payment methods do you accept?
Cards are supported for monthly and annual plans. Crypto is supported for 12-month prepay, with no auto-renewal. Annual prepay saves 20%.
Data, security and residency
Where does my data go?
You choose the deployment: in-region cloud, a private VPC endpoint, or fully on-prem and air-gapped for regulated workloads. The data residency guide explains what to verify in the contract, including subprocessors and transfer mechanisms.
Do you train on my prompts or documents?
No. The privacy policy states that Plugsky does not use prompts, completions or uploaded documents to train any AI model. You can export and delete your data.
How is data encrypted?
Data at rest uses AES-256 with envelope encryption and per-region KMS keys. Traffic uses TLS 1.3 with HSTS. Enterprise customers can use BYOK through AWS KMS, Azure Key Vault, HashiCorp Vault or an on-prem HSM. See security.
Are you SOC 2 or ISO certified?
Plugsky documents a SOC 2 Type II program covering security, availability and confidentiality, plus ISO 27001, ISO 27017 and ISO 27018 programs. Reports are shared with customers under NDA rather than published. Request the current report, period and scope through your account team.
Do you support SSO, SCIM and RBAC?
SAML/OIDC SSO, SCIM provisioning and BYOK are Enterprise controls. RBAC and audit logs are available more broadly. Audit logs can be exported to a SIEM. See security for the control list.
Do you provide a DPA and subprocessor list?
Ask for the current DPA and subprocessor list during procurement. They are part of the contract and legal review, not a marketing page. Confirm the data categories, locations and transfer mechanism against your own obligations.
API, limits and migration
What are the API rate limits?
Starter is 240 RPM with 6 concurrent requests and 2,000 requests/day. Builder is 600 RPM with 10 concurrent requests. Scale is 1,200 RPM with 16 concurrent requests. Hobby is 120 RPM with 4 concurrent requests and 1,000 requests/day. The pricing table is authoritative.
Does the API support streaming, tool calling and JSON mode?
Yes for the OpenAI-compatible chat surface: streaming, tool calling and JSON mode are supported. Model-specific support can differ, so check the model page or docs before relying on a feature in production.
How do I migrate from OpenAI?
Change the base URL to https://api.plugsky.com/v1, replace the API key, and test one chat completion, one streaming call and one tool call. The OpenAI migration checker lists the common incompatibilities.
What uptime SLA do you offer?
Paid plans are advertised with a 99.9% uptime SLA. Live availability and incident history are on the status page.
Support and deployment
How do I contact support?
Use the contact page for support and sales. Enterprise and procurement questions can go to enterprise@plugsky.com.
Do you offer VPC, on-prem or air-gapped deployment?
Yes. Private deployment options include VPC, on-prem and air-gapped environments. Start with the contact page so the team can scope the region, identity and network requirements.
Where can I see incidents and maintenance?
The status page lists current availability and incident history. For a contractual review, ask for the SLA, the measurement method and the credit process.
Prices, caps and model limits change. The pricing page, model pages and your contract are authoritative. This page is a summary, not legal advice.
Get started in minutes
OpenAI-compatible API with 39 models, a 7-day money-back guarantee, and a 99.9% uptime SLA. No code changes required.
Start Free → Read the docs