plinth

Talk to us

Start free

plinth

Talk to us

Start free

Getting started

Quickstart

Authentication

Your first route

Errors

Inference

Chat

Streaming

Structured output

Batch

Failover

Evals

Concepts

Writing a suite

Graders

CI integration

Observability

Traces

Metadata

Cost attribution

Export

Reference

API reference

SDKs

Rate limits

Changelog

Getting started

Quickstart

Authentication

Your first route

Errors

Inference

Chat

Streaming

Structured output

Batch

Failover

Evals

Concepts

Writing a suite

Graders

CI integration

Observability

Traces

Metadata

Cost attribution

Export

Reference

API reference

SDKs

Rate limits

Changelog

This page is a layout, not documentation.

Replace it with your own before publishing.

Quickstart

Your first call takes two minutes.

You need an API key and a terminal. You do not need a provider account. Routing to our default models works on the free tier from the first request.

Step 01 · Install

from plinth import Plinth

client = Plinth()
print(client.chat.create(model="plinth-route-v2").output_text)

Step 02 · Set your key

Create a key in the dashboard under Settings → API keys. Keys are scoped to a project and rotate without downtime.

from plinth import Plinth

client = Plinth()
print(client.chat.create(model="plinth-route-v2").output_text)

Step 03 · Make a call

from plinth import Plinth

client = Plinth()
print(client.chat.create(model="plinth-route-v2").output_text)

Step 04 · Look at the trace

Open the dashboard. The call you just made is already there, with the model the router picked, the token counts, and what it cost. Every call gets one, on every plan.

Rate limits are per project

Rate limits are per project, not per key. Creating more keys will not raise your limit.

Pinning a model disables failover

If that provider goes down, your request fails.

Next steps

Attach an eval suite

Set a routing policy

Add metadata to traces