Quickstart

Four steps: sign in, create a key, add an upstream, point your client at Prism.

1. Sign in

Sign in with your TYO account at prism.tyo.com.au — no card needed. You land on a free beta tenant automatically, with a 1,000 request/month quota.

2. Create an API key

From the dashboard, open API keys and click New key. Keys look like tyr-… and are shown to you once — copy it somewhere safe immediately. If you lose it, revoke it and create a new one; Prism never displays a key again after creation.

3. Add an upstream (BYOK)

Prism doesn't supply model access itself — it routes to models using your provider keys. Go to Upstreams and add at least one:

• Pick a preset (deepseek, qwen, google, kimi, glm, openai, anthropic, or ollama for a self-hosted endpoint) to prefill the base URL and default model, or choose Custom for any OpenAI-compatible URL.
• Paste your provider's API key. It's encrypted (AES-256-GCM) the moment it's saved and no part of it — not even a prefix — is ever shown again; tell your upstreams apart by the name you gave each one.
• Set a tier (1 = cheap, 2 = mid, 3 = strong) so auto routing knows where this upstream fits.

4. Point your client at Prism

The routing endpoint is https://prism-api.tyo.com.au/v1 (beta — expect occasional maintenance windows). It's OpenAI-compatible, so any OpenAI SDK or curl works by changing the base URL and key.

curl:

curl https://prism-api.tyo.com.au/v1/chat/completions \
  -H "Authorization: Bearer tyr-your-key-here" \
  -H "Content-Type: application/json" \
  -d '{"model": "auto", "messages": [{"role": "user", "content": "Hello!"}]}'

Python (openai SDK):

import openai

client = openai.OpenAI(
    base_url="https://prism-api.tyo.com.au/v1",
    api_key="tyr-your-key-here",
)

resp = client.chat.completions.create(
    model="auto",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)

model: "auto" lets Prism pick the cheapest upstream that can handle the request; use the name you gave an upstream instead for an explicit pick. Model choices, in short: "auto" (smart routing), "cascade" (strict layer order), or an upstream's name. Check your dashboard afterwards — you'll see the request, its cost, and what it saved you. See Smart routing for how the routing decision works.