Private playground

Choose, paste, ask. Nothing else to learn first.

The setup unlocks one step at a time. Your key and prompt travel from this browser to the selected provider—not through a FreeToken server proxy.

Fast pathProvider → model → key → prompt

Need a key? The correct official application guide appears after provider selection.

01

Provider and model

Pick the access route first. Only compatible models appear.

Best forGeneral chat
AccessOfficial zero-priced route
ProtocolOpenAI compatible
02

Your API key

Used only for the direct request from this browser.

Get key →
Use a saved keyChecking your encrypted Vault…
Browser → OpenRouter. No FreeToken proxy.
03 · Test one real request

Google: Lyria 3 Clip Preview

Ready
Choose a provider and model. The response will appear here.
Advanced settings
After the test

Use this route in your AI tool.

Configuration now stays beside the working model instead of living on a separate page.

Choose the target tool

Secrets are placeholders by default and never embedded in shared URLs.

Selected routeOpenRouter · Google: Lyria 3 Clip Preview
# Direct Codex config intentionally not emitted
# Selected route: OpenRouter · google/lyria-3-clip-preview
# Catalog protocol: OpenAI compatible
#
# Current Codex custom providers use the Responses API wire protocol.
# FreeToken has not verified a Responses-compatible Codex endpoint for this
# route, so setting OPENAI_BASE_URL here would create a likely-broken config.
#
# Safe choices:
# 1. Select OpenCode above for a direct provider config, or
# 2. Put OpenRouter behind a trusted local Responses-compatible adapter,
#    keep <YOUR_OPENROUTER_API_KEY> inside that adapter, and configure Codex to use the
#    adapter's documented endpoint.
#
# Your provider key can stay in a shell-scoped variable:
$env:OPENROUTER_API_KEY="<YOUR_OPENROUTER_API_KEY>"
1

Choose a route

Provider selection limits the model list automatically.

2

Use your key

Paste it here or hand it off temporarily from encrypted Vault.

3

Keep moving

Compare the result or export the same model into your AI tool.

Compare model details