Pricing

The Pricing section of Settings holds a catalog of per-model token prices. With prices set, Chatty estimates the cost of every model call and surfaces it on your runs and trials. The catalog is managed by the PricingManager component.

Use pricing when you care about the cost of running agents and experiments — especially for multi-agent panels and large trial sweeps, where token usage adds up quickly.

The pricing catalog

For each model you record:

Field	Meaning
Input price	Cost per input (prompt) token, usually per 1K or 1M tokens
Output price	Cost per output (completion) token

Prices are defined per model, so different models — and different providers — can each carry their own rates.

Setting a price

1. Open Settings and go to Pricing.

2. Add or edit a model entry and enter its input and output token prices.

3. Save. New cost estimates use these prices immediately.

Where cost surfaces

Once prices are set, cost is estimated from token usage and shown on:

• Chat and graph runs — the cost of the model calls in a run.
• Trials — cost per trial, so you can compare configurations by cost as well as quality.

Note

Cost is a first-class signal for experiments. Set prices for every model you run so trial comparisons include cost alongside accuracy and latency.

Related settings

Providers & API keys Connect the providers whose models you price here.