Prices and comparison
What a role-play message costs on this service and how that sits next to other options. All prices are in US dollars and are shown as of 2026-11-02. Prices change: this page shows the prices in force on that date.
Adults only (18+). See the acceptable use rules.
Our prices
You pay per token from your balance, and tokens the model spends on hidden reasoning are billed like any other output. This service has two lines, Standard and Super Deal (economical tier); the figures on this page are for Super Deal, and the prices of both lines are on the pricing page. Super Deal has no backup route outside its own upstream network.
| Line | Price per million tokens | Cost of a thousand role-play messages |
|---|---|---|
| Super Deal (economical tier) | 0.028 USD per 1 million tokens, same for input and output | measured 0.2315 USD |
How we measured the Super Deal (economical tier) figure: 30 requests (15 Russian and 15 English), average input 8062 tokens, average output 182 tokens, output limit 400 tokens, with the amount read from the account balance. Single requests ranged from 0.2181 to 0.2559 per thousand messages. The figure is an upper estimate: other processes shared the same balance.
What changes the price
The cost depends mostly on how long the chat history is. Calculated from tokens at the Super Deal (economical tier) rate, for a thousand messages with an output limit of 400 tokens:
| Input per message | Cost of a thousand messages |
|---|---|
| 2000 to 3999 tokens | 0.090 USD |
| 4000 to 5999 tokens | 0.146 USD |
| 6000 to 7999 tokens | 0.206 USD |
| 8000 tokens and more | 0.238 USD |
Across a mixed test of 220 requests the billed average was 0.158 USD per thousand messages. The client setting matters little: against the regular model at a limit of 400, the fast alias is about 0.9 percent lower and the regular model at a limit of 1000 about 3.1 percent higher, in all cases below 0.01 USD per thousand messages. See the connection page for the settings.
Next to the market
The comparison uses the same message shape for every row: 8062 input and 182 output tokens, converted to the cost of a thousand messages. Sources were public pages read on 2 November 2026 unless a row says otherwise. Rows marked secondary come from articles, not from the seller's own page.
| Option | Cost of a thousand messages | Note |
|---|---|---|
| This service, Super Deal (economical tier) | 0.2315 USD | measured, see above |
| Direct DeepSeek API, off-peak, no cache | 1.32 USD | list rates 0.15 USD input and 0.60 USD output per 1 million tokens, from the DeepSeek pricing page |
| Direct DeepSeek API, peak hours, no cache | 2.64 USD | list rates 0.30 USD input and 1.20 USD output per million tokens; peak hours are on weekdays |
| Direct DeepSeek API, off-peak, if 90 percent of the input comes from cache (assumption) | 0.25 USD | we have not measured the cache share in role-play chats; with a high share the direct price comes close to ours |
| OpenRouter, cheapest listed host for DeepSeek-V4-Flash-0731 | 0.40 USD | one reading of the OpenRouter model page; host prices change |
| Chutes (proxy used with Janitor AI), DeepSeek 3.2, after its price increase | 8.24 USD | secondary source (figure as stated in the article), an article dated 2026-06-20 and updated 2026-09-19; a different model |
Notes on the table. The direct DeepSeek page lists a newer model version than the one we run, so treat those rows as a market reference, not as the same model. Per-token prices are only one part of a choice: we make no claim about the quality of the text.
Flat subscriptions
A flat plan can be cheaper than paying per token if you use all of it, and more expensive if you do not. The per-message figures assume 3000 messages used, which is our assumption, not a measurement.
| Option | Plan | Cost of a thousand messages |
|---|---|---|
| NanoGPT (proxy service), secondary source | 8 USD with about 60,000 requests | 0.13 USD if the whole quota is used, 2.67 USD at 3000 messages |
| Character.AI c.ai+, secondary source | 9.99 USD (94.99 USD for a year) | 3.33 USD at 3000 messages |
| This service, pay per token | no subscription | for 3000 messages: Super Deal (economical tier) 0.69 USD |
The subscription rows come from articles dated 2026-06-08 and 2026-06-20 and may be out of date; the NanoGPT quota counts requests regardless of context size, and we have not checked that. A subscription and per-token billing are not the same product, so this comparison is only indicative.
Next
How to connect: connection instructions. What we measured about delivery and where it falls short: reliability. Rules: acceptable use, terms of service.