Billing Rules

Glossary

Beans

The virtual currency of this platform, the fees generated by platform chats are paid by beans.

Token

Token is the billing unit in the chat model. One Chinese character is approximately equal to 1.4-2.0 tokens, and one English word is approximately equal to 1.3 tokens.

Reminder: User input, historical messages, and AI replies all consume tokens. Deleting irrelevant historical messages in the chat or setting the maximum number of messages sent can reduce the consumption of tokens, but historical messages are very useful for AI to understand the context and answer.

Basic text conversation

For general use scenarios, it is recommended to use Xiaomo, which is sufficient. Claude, Gemini and other cost-effective robots are also highly recommended. GPT-4 has a high fee and is recommended for scientific research or professional purposes.

Name	Fee	Maximum Token Amount	Description
Xiaomo/GPT-4o mini (GPT-4o-mini)	0.6 Beans/Token	128000	Default version
Xiaomo/GPT-3.5 16k (gpt-3.5-turbo)	0.75 Beans/Token	16385	Reserved old version
GPT-4 4o (gpt-4o)	7.5 Beans/Token	128000	The latest, cheaper, faster version, the dataset ends in 202310
GPT-4 Turbo 128k (gpt-4-turbo)	15 Beans/Token	128000	Old version retained, the dataset ends at 202312
Claude3 (claude-3-haiku)	0.5 Beans/Token	200000
Claude3 (claude-3.5-sonnet)	6 Beans/Token	200000
Claude3 (claude-3-opus)	30 Beans/Token	200000
Claude3.5 (claude-3-5-sonnet)	6 Beans/Token	200000
Wen Xin Yi Yan (ERNIE-4.0-8K)	8 Beans/Token	8192
Wen Xin Yi Yan (ERNIE-3.5-8K)	0.8 Beans/Token	8192
Wen Xin Yi Yan (ERNIE-Lite-8K)	0.1 Beans/Token	8192
Tong Yi Qian Wen (qwen-long)	0.2 Beans/Token	200000+
Tong Yi Qian Wen (qwen-turbo)	0.4 Beans/Token	8192
Tong Yi Qian Wen (qwen-max)	8 Beans/Token	8192
Tong Yi Qian Wen (qwen2-7b-instruct)	0.2 Beans/Token	8192
Tong Yi Qian Wen (qwen/qwen2-72b-instruct)	0.7 Beans/Token	128000
ChatGLM (glm-4)	7 Beans/Token	128000
ChatGLM (glm-3-turbo)	0.1 Bean/Token	128000
Gemini (gemini-1.5-flash)	0.1 Bean/Token 0.35 Bean/Token	Input 1048576 Output 8192	Image recognition 1000 Token (100 Beans)/conversation
Doubao (Doubao-pro-128k)	0.6 Bean/Token	128000
Doubao (Doubao-lite-128k)	0.1 Bean/Token	128000
Kimi (moonshot-v1-8k)	0.8 Beans/Token	8192
Kimi (moonshot-v1-32k)	1.6 Beans/Token	32768
Kimi (moonshot-v1-128k)	4 Beans/Token	128000
Tencent Hunyuan (hunyuan-lite)	0.1 Beans/Token	Maximum input 250k Maximum output 6k
Tencent Hunyuan (hunyuan-standard)	4 Beans/Token	Maximum input 250k Maximum output 6k
Tencent Hunyuan (hunyuan-pro)	7 Beans/Token	Maximum input 28k Maximum output 4k
Llama3 (llama3 8B)	Free for a limited time	8192
MistralAI (mistralai/mistral-7b-instruct)	Free for a limited time	4000

After calculation, the part less than 1 bean will be rounded up.

Extended functions

Text generation image

Painting Master (Stable Diffusion version), 10% discount:

• 256*256px: 8000 beans/sheet 800 beans/sheet

• 512*512px: 9000 beans/sheet 900 beans/sheet

• 1024*1024px: 10000 beans/sheet 1000 beans/sheet

Painting Master (DALL·E 3 version): 20000 beans/sheet 18000 beans/sheet

AI image editing

1000 beans/sheet

Image recognition

• 512 * 512: 255 Token = 3825 beans
• 1024 * 1024: 765 tokens = 11475 beans

Detailed calculation rules: First, regardless of the size of the image, 85 basic tokens will be consumed, and then the image will be divided into 512 * 512 blocks. For example, at least n 512*512 blocks are required, so the total token consumption is n * 170 + 85

Voice to text

50 beans/second

Text to sound

2 beans/character (each Chinese character, English letter, number, symbol is counted as one character)

Document parsing

0.01 beans/Token

Text and image to video

20000 beans/video

Plugins

The operation of plugins requires additional prompts and multiple calls to complete the entire process. Compared with not running plugins, the consumption of tokens will increase by about 2 times, and the plugin call fee needs to be paid. If you just turn on the plugin and do not call it, these fees will not be incurred. Whether to call the plugin is determined by the robot intelligence.

The call charges for plugins are as follows:

• DALL·E 3: 20000 beans/slot 18000 beans/slot

• Search engine: free

• Document parsing: 0.01 beans/token

Get beans

Purchase

• Recharge and redeem, log in to the recharge platform to recharge beans, the current price is 0.000028 yuan/bean-0.0000375 yuan/bean

Reminder: Based on the general usage scenario of Xiaomo (gpt-3.5), 30 yuan is enough for one month of use, or even longer. Based on gpt-4, the consumption will be 20-40 times

Free

• New users will be given 1,000 beans for initial experience

• Daily collection, 300 beans can be collected daily, through the menu [Avatar] -> [Daily Free Beans]

• Publish a robot, create a robot in Bumo Assistant and review and publish it, and you will get 100,000 beans

• Recommend friends to register, enter Invite Gift, get the referral link, invite friends to register and use it, the referrer will get 10,000 beans, and the referee will get 5,000 beans

Consumption records can be viewed in Consumption Flow, and consumption within 10 minutes will be merged into one record; Recharge and free acquisition records can be found in Fund Recharge .

(The above pricing will be adjusted according to operating costs)