Connect multiple AI modelsthrough one API.
Use supported GPT, Claude, Gemini, and other models with the same API key. Add KRW credit and review usage and cost by model.
Set the API URL, key, and model for your SDK.
from openai import OpenAI
client = OpenAI(
base_url="https://apirouter.pleum.ai/v1",
api_key="plm_...",
)
resp = client.chat.completions.create(
model="gpt-4.1",
messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)Set the boundaries for your team's AI use.
Set limits and allowed models for each API key.
Separate usage conditions by key, then review team usage and costs.
Review model rates and payment costs.
Model rates
By model
Check the latest rate for each model you plan to use.
Top-up and conversion
Terms apply
Review top-up fees and currency conversion terms before payment.
Tax and payment
By payment
Check VAT and payment fee terms on the billing screen and in the current guide.
Know how your requests are handled.
Data processing
Requests are sent to the AI provider you select and may be processed outside Korea. Optional personal-data masking has a limited scope.
Failures and fallback
Requests that fail before a response are not charged. Partial streaming responses may be billed for actual usage. Fallback is available only on supported routes.
Check these details before you start.
- What needs to change in my code?
- Check the API URL, API key, and model in your SDK configuration. Some SDK and model features require additional settings; see the setup guide.
- How do I get a tax invoice?
- Review the current requirements, including business verification and an invoice request. Invoices are not described as automatic.
- How is usage billed?
- Model rates, top-up and conversion fees, VAT, and payment fees may apply. Check the current pricing page and billing screen.
- What happens when a request fails?
- Requests that fail before a response are not charged. A partially delivered stream may be billed for actual usage. Fallback works only for supported requests.
Connect a model.Then review usage.
Check supported models and current pricing before creating an API key.