Playground

Test ORGN Gateway models interactively from the console Playground. Select TEE or ZDR models, send prompts, and inspect responses before integrating into your application.

The Playground in the Gateway console lets you send inference requests directly from the browser. Use it to evaluate models, compare TEE and ZDR behavior, and validate prompts before wiring a model into your application.

The Playground uses your console session (id-orgn SSO) — it does not require a separate API key for interactive testing. For production integrations, create an sk-ollm-* key and call https://api.gateway.orgn.com/v1 directly.

Opening the Playground

Open gateway.orgn.com and authenticate with id-orgn.

Navigate to Playground

Select Playground from the console sidebar.

Select a model

Choose a model from the dropdown. The catalog includes both TEE models (near_*, phala_*) and ZDR models (vercel_*). Model IDs use underscores throughout.

Sending a request

Choose a model from the selector. The execution type (TEE or ZDR) is indicated alongside the model name.
Compose your prompt in the message input. Multi-turn conversations are supported — add assistant and user messages to simulate a dialogue.
Adjust parameters such as temperature and max tokens if exposed for the selected model.
Send and inspect the response in the output panel.

The Playground calls the same Gateway inference path as the production API. A model that works in the Playground will behave identically when called with your sk-ollm-* API key.

What the Playground shows

Data	Visible in Playground	Notes
Model output	Yes	Full completion text
Token usage	Yes	Input, output, and total counts
Latency	Yes	End-to-end request time
Model ID	Yes	Underscore format (e.g. `phala_deepseek_r1`)
Execution type	Yes	TEE or ZDR indicator
Attestation details	No	Use scanner.orgn.com for TEE verification

Attestation verification is not available in the Gateway console. After testing a TEE model in the Playground, send a production request with your API key and inspect the attestation receipt in Scanner.

TEE vs ZDR testing

Use the Playground to compare execution types before committing to an integration:

Type	Prefix examples	Playground behavior	Post-request verification
TEE	`near_qwen3_30b`, `phala_deepseek_r1`	Hardware-isolated inference	Verify attestation in Scanner
ZDR	`vercel_claude_sonnet_4_6`, `vercel_gpt_5`	Policy-enforced zero retention	No attestation receipt

Not every model is TEE. The Playground model selector surfaces execution type so you can test the right model for your compliance requirements.

From Playground to production

Once you have validated a model in the Playground:

Create an API key under API Keys.
Copy the model ID exactly as shown (underscores, not slashes).
Send the same request to the production endpoint:

curl -X POST https://api.gateway.orgn.com/v1/chat/completions \
  -H "Authorization: Bearer sk-ollm-YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "phala_deepseek_r1",
    "messages": [
      { "role": "user", "content": "Your playground prompt here" }
    ]
  }'

Verify TEE requests in scanner.orgn.com if your workload requires cryptographic proof.

Limitations

The Playground is for interactive testing, not high-volume production traffic.
Streaming behavior may differ slightly from the raw API depending on console UI implementation.
Some model-specific parameters (reasoning effort, tool calling) may require direct API access.

What's next

Model List — browse the full catalog
API Keys — create keys for production
OpenAI SDK integration — drop-in SDK migration
Model catalog — TEE vs ZDR selection guide

On this page