What does this cost me per token, roughly, and what drives the bill?
Between nothing and a few cents per call, depending entirely on the model. The app shows each model's price per million tokens next to it. What actually drives a bill is not the per-token rate — it is chart images and deep Coach questions.
The prices the app carries, in US dollars per million tokens, input and output:
- FreeEvery OpenRouter model marked (free): $0 in, $0 out.
- GroqLlama 3.1 8B at $0.05 / $0.08; Llama 4 Scout at $0.11 / $0.34; Llama 3.3 70B at $0.59 / $0.79.
- OpenAIGPT-4o mini at $0.15 / $0.60; GPT-5 mini at $0.25 / $2.00; GPT-5.1 at $1.25 / $10.00.
- AnthropicClaude Haiku 4.5 at $1.00 / $5.00; Sonnet 5 at $3.00 / $15.00; Opus 5 at $5.00 / $25.00.
Check these against your provider's own pricing page before making a decision on them — they are the figures the app estimates with, and providers change prices.
What actually costs money. Per-token rates are the least interesting part, because most calls are small. Three things dominate:
- Chart reads. An image is worth a great many tokens, and a full-screen multi-panel capture is worth more than a single chart. This is the single most expensive routine call the app makes.
- Deep Coach questions. A question that sends the Coach off to fetch your trades, your setups and your levels before answering runs several calls, each carrying context. This is what two-phase economy mode is for.
- Bulk runs. Tagging fifty trades is fifty calls. Cheap per call, visible in total, which is why both bulk actions are batched and capped.
What is nearly free: daily summaries, debrief drafts, sentiment tags, and anything routed to a free provider.
