Billing Model

Chuizi.AI charges upstream provider cost x 1.05 with no hidden fees. All billing calculations use 8-decimal precision to ensure financial accuracy.

Three Billing Types

Billing Type	Use Case	Unit
Per-token	Chat, Embedding, and text models	Per 1M tokens
Per-request	Image generation, TTS	Per request
Per-second	Speech-to-text (Whisper)	Per second of audio

Per-Token Billing

Most models (Chat, Reasoning, Embedding) use per-token billing. Input tokens and output tokens are priced separately.

Formula:

cost = (input_tokens x input_price + output_tokens x output_price) x 1.05

Example — Claude Sonnet 4-6:

Item	Count	Price (per 1M tokens)	Subtotal
Input tokens	2,000	$3.00	$0.006000
Output tokens	500	$15.00	$0.007500
Subtotal	—	—	$0.013500
x 1.05 multiplier	—	—	$0.014175

If the request uses prompt caching, the cost also includes cache_read_tokens and cache_write_tokens. See Cache Discount Pricing for details.

Token Subtypes

Some models report additional usage fields:

Field	Description	Common Models
`input_tokens`	Input token count	All models
`output_tokens`	Output token count	All models
`cache_read_tokens`	Cache-hit token count	Anthropic, OpenAI, DeepSeek
`cache_write_tokens`	Cache-write token count	Anthropic
`reasoning_tokens`	Reasoning chain token count	o3, o4-mini, DeepSeek R1

Reasoning tokens are billed at the output price and included in the output_tokens total. For details on how billing is processed after each request, see Billing Flow.

Per-Request Billing

Image generation and speech synthesis models charge a fixed price per request, regardless of token count.

Common model prices:

Model	Billing Type	Upstream Price	Actual (x1.05)
Imagen 4.0	Per request	$0.040	$0.042
DALL-E 3 (1024x1024)	Per request	$0.040	$0.042
Nova Canvas	Per request	$0.040	$0.042
GPT-4o-mini-tts	Per request	Per character	Per character x 1.05

Per-Second Billing

Speech-to-text models are billed by audio duration.

Model	Upstream Price	Actual (x1.05)
Whisper (gpt-4o-transcribe)	$0.00006/sec	$0.000063/sec

Query Model Prices

Call GET /v1/models to retrieve real-time pricing for all models:

terminal

bash

curl https://api.chuizi.ai/v1/models \
  -H "Authorization: Bearer ck-your-key"

Each model in the response includes a pricing field:

config.json

json

{
  "id": "anthropic/claude-sonnet-4-6",
  "pricing": {
    "input": "3.00",
    "output": "15.00",
    "cache_read": "0.30",
    "cache_write": "3.75",
    "unit": "per_1m_tokens"
  }
}

All prices shown are upstream costs. Your actual bill = displayed price x 1.05.

The 1.05x Multiplier

Chuizi.AI applies a flat 5% service fee covering gateway operations, multi-protocol support, smart routing, and automatic failover. Compared to managing API keys across multiple providers yourself, the 5% markup is typically more cost-effective.

See Model Pricing for the full price list.

Next Steps

Cache Discount Pricing — Save up to 90% on input costs with prompt caching
Billing Flow — How balance reservations and reconciliation work
Model Pricing — Complete price table for all supported models