# Price Transparency

Canonical page: `https://nanthai.tech/features/price-transparency`

Tier: Free

See the cost of every response, your running total, and your remaining balance — in the chat.

Advanced Stats shows per-message generation cost, a live chat total, and a 4-bucket breakdown (Responses, Memory, Search, Other). Your OpenRouter balance sits right in the header. Full visibility, zero markup.

## How it works

1. **Enable Advanced Stats** — In Settings, toggle "Show Advanced Stats". That's the only step — cost data starts appearing immediately in every chat.
2. **See the cost of every response** — Each assistant message shows its generation cost next to the timestamp — e.g. "3:45 PM $0.0023". The figure reflects only that response's tokens, not background tasks.
3. **Check the chat total** — The header subtitle updates in real time to show the full session cost: "GPT-4o $0.0158". Tap it to open the cost breakdown.
4. **Drill into the breakdown** — The popover splits costs into four buckets: Responses, Memory, Search, and Other. You see exactly which features are using your credits.

## What you can do

- **Per-message cost** — Every assistant message shows its individual generation cost — 4 decimal places, right next to the timestamp. No guessing, no rounding.
- **Live chat total** — The chat header shows a running total that updates after every response. Tap it to open the full cost breakdown without leaving the conversation.
- **4-bucket breakdown** — Costs split into Responses (your AI replies), Memory (extraction and retrieval), Search (web and paper queries), and Other (title generation, compaction). Know where every cent goes.
- **Balance in chat** — Your OpenRouter credit balance appears right in the chat header. Low? You'll see it before you run out — no surprise interruptions.
- **Multi-model cost tracking** — In multi-model chats, each model's response shows its individual cost. A group total appears at the bottom so you can compare model efficiency at a glance.
- **Zero markup — verified** — Every cost figure comes straight from OpenRouter. NanthAI adds no markup. The number you see is the number OpenRouter charged — nothing more.
- **Streaming-aware** — Costs appear as "$--" while a response is streaming and resolve to a real figure the moment the model finishes. No stale placeholders.
- **Toggle any time** — Prefer a cleaner UI? Turn Advanced Stats off and cost figures disappear. The feature is opt-in and non-intrusive — you're always in control.

## What each cost bucket includes

- **Responses** — The primary generation cost — every token your AI model produces as a reply. This is the number shown per-message.
- **Memory** — Background calls for memory extraction (finding facts to save), embedding (storing them for retrieval), and similarity search (finding relevant memories before each reply).
- **Search** — Web and academic paper searches: query generation, web search calls, research planning, source analysis, and synthesis steps in a full research pipeline.
- **Other** — Housekeeping tasks: automatic title generation, context compaction when a conversation gets very long, and subagent orchestration calls.

## Perfect for

- Developers running experiments who want to know exactly how much each model call costs.
- Cost-conscious users who want to stay on budget and avoid surprise OpenRouter charges.
- Teams using Memory and Search features who want to understand the full cost of each conversation.
- Power users comparing model efficiency — same output, different price per token.

## Related features

- Bring Your Own Key (`/features/byok`)
- Pro vs Free (`/features/pro-vs-free`)
- Memories (`/features/memories`)