Short answer. Plugsky is an OpenAI-compatible AI API operated by Plugsky Inc. of Newark, Delaware. It exposes 39 public model IDs, flat monthly plans from $8 to $120, no monthly token cap on Builder and Scale, and cloud, VPC, on-premises or air-gapped deployment.
Entity
| Fact | Value | Source (checked 20 Nov 2026) |
|---|---|---|
| Legal name | Plugsky Inc. | plugsky.com/legal/terms |
| Registered address | 131 Continental Dr, Suite 305, Newark, DE 19713, United States | plugsky.com/legal/terms |
| Product | OpenAI-compatible AI model API and deployment platform | plugsky.com/docs |
| API base | https://api.plugsky.com/v1 | api.plugsky.com/v1/models |
| Website | https://plugsky.com | plugsky.com |
| Enterprise contact | enterprise@plugsky.com | plugsky.com/contact |
Product and API
Plugsky is an API product, not an AI writing assistant. The chat endpoint is POST https://api.plugsky.com/v1/chat/completions. It follows the OpenAI request and response shape, so the OpenAI Python, Node, Go and curl clients work after a base URL, key and model ID change. Source: plugsky.com/docs, checked 20 Nov 2026.
- Live capabilities: chat completions, streaming, JSON mode, function calling, embeddings, RAG, agents, image generation and editing, and AI Video Studio. Status page, checked 20 Nov 2026.
- Listed as coming soon: audio, moderation, files, batch, fine-tuning, assistants and the Responses API. Status page, checked 20 Nov 2026.
- Model count: 39 public model IDs. The Free plan includes
plugsky-microandplugsky-lite; paid plans include the full catalogue. API catalogue, checked 20 Nov 2026. - Billing unit: flat monthly or annual subscription. There is no per-token overage on self-serve plans. Pricing, checked 20 Nov 2026.
Models and context exceptions
The API catalogue is authoritative. Main text models serve 262,144 tokens of context. The table lists the exceptions and the engine that actually serves each public ID. Legacy IDs keep their old names but are served by the engine shown; do not copy another model family’s parameter count or benchmark onto them.
| Public IDs | Served by | Context | Note |
|---|---|---|---|
| frontier, pro, max, ultra, reasoning, thinker, kimi, longctx, coder, coder-ultra, deepseek-pro, llama4 | NVIDIA Nemotron 3 Ultra 550B MoE (A55B) | 262,144 | Kimi, DeepSeek and Llama IDs are legacy compatibility IDs. |
| micro, lite, plus, nano, mistral-small, phi, tiny, nano-omni, gpt-oss-20b, lightning, parse, coder-fast, qwen-next, gemma3-nano-2b, gemma3-nano-4b, deepseek-flash, gpt-oss, minimax | NVIDIA Nemotron 3 Nano Omni 30B (A3B) | 262,144 | Some legacy IDs map here rather than to the family in their name; the API description lists 30B (activated ~3B). |
| mistral-medium | Mistral Large 4 1T (A52B) | 131,072 | Distinct Mistral engine. |
| vision-fast, qwen-vl, gemma-4 | NVIDIA Nemotron 3.5 Super VL 120B (A12B) | 131,072 | Vision and document workloads; Qwen and Gemma names are legacy IDs. |
| translate | NVIDIA Riva Translate 4B | 32,768 | Translation specialist. |
| gpu-27b | GPU Share route | 32,768 | Actual engine is returned in x-plugsky-served-model. |
| embed, embed-nim, embed-multilingual | NVIDIA Nemotron 3 Embed 1B | 8,192 | 2,048-dimension embeddings. |
Sources: API catalogue and model pages, checked 20 Nov 2026.
Plans and published limits
| Plan | Price | Requests/day | Tokens/day | Concurrent | RPM | Input / output | Monthly token cap |
|---|---|---|---|---|---|---|---|
| Free (new) | $0 | 100 | 50,000 | 1 | 30 | 32K / 2K | Daily cap only |
| Free (legacy) | $0 | 500 | 250,000 | 2 | 30 | 32K / 2K | Daily cap only |
| Hobby | $8/mo | 1,000 | 2M | 4 | 120 | 64K / 8K | Daily cap only |
| Starter | $20/mo | 2,000 | 5M | 6 | 240 | 128K / 16K | Daily cap only |
| Builder | $60/mo | Fair use | Fair use | 10 | 600 | 256K / 32K | None |
| Scale | $120/mo | Fair use | Fair use | 16 | 1,200 | 512K / 32K | None |
Annual billing is 20% off. Paid plans include a 7-day money-back guarantee. New monthly customers get 30% off the first month, capped at $20. Free caps apply to accounts created from 2 November 2026; older accounts keep the previous allowance until migration. Sources: plugsky.com/pricing and /features/unlimited-api, checked 20 Nov 2026.
What “unlimited” means
Unlimited means no monthly token cap and no per-token overage, not infinite capacity. Builder and Scale do not meter monthly tokens; they still enforce fair-use concurrency, requests per minute, per-request context and maximum output. Hobby and Starter publish daily request and token caps instead of a monthly cap. Source: plugsky.com/features/unlimited-api, checked 20 Nov 2026.
Data regions
| Region | Code | Location | Availability |
|---|---|---|---|
| United States | us-east-1 | Virginia | Self-serve |
| European Union | eu-central-1 | Frankfurt | Self-serve |
| GCC | me-central-1 | United Arab Emirates | Self-serve |
| APAC | ap-southeast-1 | Singapore | Self-serve |
| Riyadh | — | Saudi Arabia | Enterprise only; not a self-serve default |
For a country without a live region, the options are the nearest live region, a private VPC endpoint, on-premises deployment or air-gapped deployment. Those are different commitments; do not describe them as a local data centre. Sources: docs and data residency guide, checked 20 Nov 2026.
Deployment
- Plugsky cloud: shared multi-tenant service in the live regions above.
- Private VPC or endpoint: dedicated network boundary, contract-led.
- On-premises: software deployed in the customer’s environment, contract-led.
- Air-gapped: isolated deployment for restricted networks, contract-led.
Private deployment scope, hardware requirements and support terms are confirmed in writing. Do not assume a private deployment is self-serve. Source: plugsky.com/docs, checked 20 Nov 2026.
DPA, SOC 2, HIPAA and SLA status
| Item | Status | What to request |
|---|---|---|
| DPA | GDPR Article 28 terms and EU SCCs; PDPL, DIFC and NSD alignment documented. Available on paid plans; bespoke addenda on Enterprise. | Signed DPA and subprocessor list |
| SOC 2 | SOC 2 Type II report available under NDA; confirm the current scope and audit period. This page is not an attestation and does not itself certify Plugsky. | Current report under NDA; verify the period and exceptions |
| HIPAA | BAA available by contract, subject to review. Not a HIPAA certification. | BAA and scope review |
| SLA | 99.9% uptime target. The contract schedule governs; Hobby is not in the SLA table. | Plan-level SLA and exclusions |
| Other certifications | No other certification is claimed on this page. No government authorization is claimed. | Current status in writing |
Sources: DPA, security page and pricing, checked 20 Nov 2026. Contract terms govern over summary pages.
Contact
Enterprise and procurement: enterprise@plugsky.com. General contact: plugsky.com/contact. Service status: plugsky.com/status. For plan and model facts, cite /pricing and /models rather than article copy.
Frequently asked questions
How many models does Plugsky offer?
Plugsky exposes 39 public model IDs through one API. The Free plan includes plugsky-micro and plugsky-lite; paid plans include the full catalogue. Main text models support 262,144 tokens of serving context, with the exceptions listed on this page.
What is the Plugsky API endpoint?
The chat completions endpoint is POST https://api.plugsky.com/v1/chat/completions. It is OpenAI-compatible, so existing OpenAI SDK clients work by changing the base URL, API key and model ID.
Which Plugsky plans have no monthly token cap?
Builder at $60 per month and Scale at $120 per month have no monthly token cap and no per-token overage. Builder allows 10 concurrent requests and 600 RPM; Scale allows 16 concurrent requests and 1,200 RPM. Fair-use limits and per-request context ceilings still apply.
Where does Plugsky process customer data?
Live regions are US East (us-east-1, Virginia), EU (eu-central-1, Frankfurt), GCC (me-central-1, UAE) and APAC (ap-southeast-1, Singapore). Riyadh is available on Enterprise and is not a self-serve default region. Customers outside those regions can use the nearest live region, a private VPC, on-premises or air-gapped deployment.
What is Plugsky's SOC 2 status?
Plugsky publishes a SOC 2 Type II report under NDA. Confirm the audit period, scope and any exceptions during vendor review; this page is not the attestation and does not itself certify Plugsky.
Related canonical pages
Flat-rate AI API · Pricing · Unlimited API definition · Model catalogue · Documentation · Security procurement
Sources
- plugsky.com/pricing — plans, caps, discounts; checked 20 Nov 2026.
- plugsky.com/features/unlimited-api — unlimited definition; checked 20 Nov 2026.
- api.plugsky.com/v1/models — public model IDs and context; checked 20 Nov 2026.
- plugsky.com/docs — API, deployment and regions; checked 20 Nov 2026.
- plugsky.com/legal/dpa — DPA terms; checked 20 Nov 2026.
- plugsky.com/security — security program; checked 20 Nov 2026.
Last checked 20 November 2026. If a figure here differs from the pricing page or API response, the pricing page and API win.