> ## Documentation Index
> Fetch the complete documentation index at: https://www.plungeai.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# PlungeAI API Pricing

> How PlungeAI bills model calls: the wallet, the three prices, and the posted price list.


PlungeAI bills model calls from your prepaid wallet, per token, at one of three prices. The numbers below come from `GET /v1/pricing` when these docs are published; the [live table](https://plungeai.com/pricing/) updates over a WebSocket the moment a price changes.

## How billing works

- **Wallet, reserve, settle.** Before the upstream call the gateway holds a conservative maximum from your wallet; after it, the hold settles at the true price and the rest is released. A failed call releases the hold: no charge. A near-empty wallet answers `402 insufficient_balance`.
- **What you were charged** is in the response: `usage.cost` and the `x-ocean-cost-usd` header carry the price you pay (never the provider cost). `GET /v1/generation?id=…` returns it afterwards, with `pricing_mode`.
- **Trial and active accounts.** Every request is metered; only `active` accounts are debited. A `trial` account is priced and logged but never debited, within a $10 a month platform-cost allowance (then `429 insufficient_quota`). An `active` key over its monthly limit answers `429 insufficient_quota`.

## The three prices

| Mode | What you pay | `x-ocean-pricing` |
|---|---|---|
| Posted | Ocean's published price for the model, the same whichever host serves it | `posted` |
| Passthrough | The provider cost times your organization markup | `passthrough` |
| BYOK | Your own provider key pays the provider; you pay a flat routing fee | `byok` |

Every completion carries `x-ocean-pricing` and `x-ocean-provider` (the host that served it). A posted model routes to the cheapest host that holds a key; the price does not change with the host.

**The promise:** a posted price never exceeds the model maker's own API list price.

**Price changes** reach your bill within 120 seconds of being published. **Aborted streams** settle the tokens that streamed, at the posted price.

## Posted prices

Per 1M tokens, USD, published 2026-10-29 01:00 UTC (`published_at` 1793235642982). List = the model maker's own API price. [See all models, live](https://plungeai.com/pricing/).

| Model | List in / out | Ocean in / out | Off |
|---|---|---|---|
| meta-llama/llama-3.3-70b-instruct | $1.04 / $1.04 | $0.125 / $0.40 | 75 % off |
| openai/gpt-oss-120b | $0.15 / $0.60 | $0.0463 / $0.2125 | 65 % off |
| deepseek/deepseek-v4.1-flash | $0.30 / $1.20 | $0.25 / $0.75 | 33 % off |

Every model, including passthrough ones: [Model Catalog](/models/catalog).