TalePath AI Playground

Compare model flavour, latency, format discipline, and estimated cost.

Admin dashboard

Model lab

One character. One quest. Every model side by side.

Use this to compare the models we can currently call without changing production routing. The test context is intentionally polished: a relationship-aware character, a fantasy mystery quest, and a strict 3-choice button contract.

Choose the configured admin Google account to load the playground.

Loading...

Loading...

Run comparison

Tip: run one surface at a time. Quest choices are the fastest way to expose weak instruction following.

Text generators

Curated for TalePath: affordable, non-free, general-purpose text models with enough context for chat, narration, and 3-choice JSON discipline.

Results

Run a model to see output here.

Image generation

Current app generators and low-cost image candidates live here, away from the text model picker.

No character reference selected.

Ready to test the current prompt across the carousel styles.

Run all picker styles to compare the real generator output.