Pricing

In one line

Billed by actual token usage in real time, top up ¥7.14 to get $1 in platform credit (fixed rate). No monthly fee, no minimum spend, no plan lock-in. Your balance never expires, and unused credit is refundable.

Price table

See Models. All prices are in RMB per million tokens (¥/MTok).

Prices are adjusted promptly as upstream costs change. Changes are announced on the main site beforehand, and calls already made are settled at the price in effect at the time.

Billing logic

Each call is billed on input tokens + output tokens separately:

charge = input tokens × input rate + output tokens × output rate
                                      (rates in ¥/MTok)

For example, using gpt-5.5 on a 2,000-character Chinese article (~3,000 tokens) and asking for a 500-character summary (~800 tokens):

charge = 3000 × 2.5 / 1,000,000  +  800 × 15 / 1,000,000
       = 0.0075 + 0.012
       ≈ ¥0.02

About two cents. The rates above are for illustration only; actual pricing is shown in the console.

About tokens

Chinese is roughly 1 character ≈ 1.5–2 tokens; English roughly 1 word ≈ 1.3 tokens; code and symbols fragment more than plain text. Each model uses its own tokenizer: estimate GPT models with OpenAI's Tokenizer; Claude and Gemini use different tokenization rules — see Anthropic's docs and Gemini's docs for their respective counting methods.

Top-ups

• Minimum ¥1, no cap.
• Currently supports WeChat Pay, Alipay, and corporate bank transfer.
• Credit lands immediately after payment.
• Invoices are supported — contact support to request one.

Balance and refunds

• Balance never expires and is never cleared even if the account is inactive.
• Unused balance is refundable to the original payment channel.
• Consumed amounts are accounted from call records, viewable line by line on the console "Usage" page.

Viewing your bill

The console "Usage" page provides:

• Daily / weekly / monthly spend curves
• Breakdown by model / by API Key
• Per-call token count, charge, and status code

You can also fetch usage programmatically:

curl https://api.tokone.ai/v1/usage \
  -H "Authorization: Bearer $TOKONE_API_KEY"

If a charge looks wrong, contact support and it'll be reconciled against the logs line by line.

Platform fees

The platform charges no extra fees — no "subscription fee," "service fee," "processing fee," or "minimum spend." The number on your bill is the model's token rate × usage. Our revenue is the margin between upstream cost and the listed price, which you can sanity-check directly against the official APIs.