How do you build with Plugsky in 5 minutes?

Key facts

Chat endpoint	POST https://api.plugsky.com/v1/chat/completions
SDK	pip install openai — the official OpenAI SDK works unchanged
Base URL	https://api.plugsky.com/v1
Auth	Authorization: Bearer sk-live-…
Free tier	2 free AI models (plugsky-micro, plugsky-lite), no card required
Models	30+ models; start with plugsky-lite and scale to plugsky-pro
Live capabilities	Streaming, JSON mode, function calling, embeddings and vision on supported models
Roadmap endpoints	Audio, images, moderation, files, batch, fine-tuning, assistants and responses are coming soon

TL;DR

• First call takes about five minutes: key, SDK, base URL, request.
• The OpenAI SDK works as-is — only the base URL and model name change.
• plugsky-micro and plugsky-lite are free to start; no card required.
• Streaming, JSON mode and tools are live on the chat endpoint.
• Verify with GET /v1/models, then design for 429 Retry-After responses.

How it works, step by step

1. Create a Plugsky account and generate an API key in the dashboard.
2. Install the OpenAI SDK: pip install openai (or npm install openai).
3. Set PLUGSKY_API_KEY in your environment and base_url to https://api.plugsky.com/v1.
4. Call client.chat.completions.create with model plugsky-lite and a short prompt.
5. Add stream=True and print delta chunks to confirm streaming works.
6. Switch to plugsky-pro for a harder prompt and compare quality.
7. List models with GET /v1/models and wire error handling for 401, 403 and 429.

Original data

Try it yourself

Open the OpenAI-compatible API tester →

Minute 0-1: create a key

Sign in to the Plugsky dashboard, open API keys and create a key scoped to your project. Copy it once — the dashboard will not show the secret again. Store it as an environment variable:

export PLUGSKY_API_KEY="sk-live-…"

The free plan issues two API keys and includes two free models, plugsky-micro and plugsky-lite, with no credit card. If you prefer a browser check first, paste the key into the OpenAI-compatible API tester before writing any code.

Minute 1-2: your first request

Install the OpenAI SDK and point it at Plugsky. Nothing else changes:

pip install openai

from openai import OpenAI

client = OpenAI(
    api_key="sk-live-…",
    base_url="https://api.plugsky.com/v1",
)

resp = client.chat.completions.create(
    model="plugsky-lite",
    messages=[{"role": "user", "content": "In one sentence: what is Plugsky?"}],
)
print(resp.choices[0].message.content)

Swap plugsky-lite for plugsky-micro for the cheapest classification-style work, or plugsky-pro when answer quality matters more than speed.

Minutes 2-3: streaming and JSON mode

Streaming is a flag, not a rewrite. Set stream=True and iterate over chunks:

stream = client.chat.completions.create(
    model="plugsky-micro",
    messages=[{"role": "user", "content": "Count to five."}],
    stream=True,
)
for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="")

For structured output, request JSON and validate it in your app:

resp = client.chat.completions.create(
    model="plugsky-pro",
    messages=[{"role": "user", "content": "Return JSON with keys ok and reason."}],
    response_format={"type": "json_object"},
)

Function calling uses the standard tools and tool_choice parameters on the same endpoint.

Minutes 4-5: verify and know the limits

List the live catalogue before you hard-code a model id:

curl https://api.plugsky.com/v1/models \
  -H "Authorization: Bearer $PLUGSKY_API_KEY"

Then plan for the error paths you will actually see: 401 for a bad key, 403 for a key missing a scope, 429 for fair-use rate limits with a Retry-After header, and 413 when a request body exceeds 16 MB. All errors use the OpenAI error schema. Endpoints such as audio, images, files, batch and fine-tuning are coming soon, so build chat, embeddings and tool workflows first and check the status page before designing around roadmap items.

Honest comparison

Capability	Plugsky	Self-hosting an OpenAI-compatible stack	Staying vendor-only
Time to first call	About five minutes	Hours to days: GPU, serving, networking	Minutes
SDK	OpenAI SDK works as-is	OpenAI SDK plus your own gateway	Vendor-specific SDK
Model access	30+ models behind one key	Only what you deploy	One vendor catalogue
Free to start	2 free models, no card	GPU cost from minute one	Vendor trial credits
Ops burden	Managed	You own GPUs, scaling and patches	Managed
Endpoint parity	Chat, embeddings and vision live; audio, images, batch and fine-tuning coming soon	Add each component yourself	Vendor-dependent

Frequently asked questions

Do I need to learn a new SDK?

No. Plugsky exposes an OpenAI-compatible /v1/chat/completions endpoint, so you keep the OpenAI SDK and change the base URL and model name.

What is the fastest way to test without code?

Open the OpenAI-compatible API tester, paste your sk-live-… key and send a chat request with plugsky-micro or plugsky-lite.

Which model should I start with?

Start with plugsky-lite for general chat, plugsky-micro for cheap high-volume tasks, and plugsky-pro when answer quality matters most.

Is the free plan enough for a prototype?

Yes. The free plan includes two free AI models and two API keys with no credit card, and a 14-day full-access trial is available.

Does streaming work with the OpenAI SDK?

Yes — set stream=True and iterate over chunks. Plugsky returns server-sent events in the same format the SDK expects.

How do I know which models exist right now?

Call GET /v1/models with your key; the catalogue page is generated from that endpoint.

What about errors and rate limits?

Errors use the OpenAI schema. Handle 401, 403 and 429; honor Retry-After on 429 and keep requests under the 16 MB body limit.

Can I use TypeScript instead of Python?

Yes. Run npm install openai, set baseURL to https://api.plugsky.com/v1 and use the same chat.completions call shape.