Magai allows you to chat with multiple AI models, each varying in complexity, speed, and quality. Behind the scenes, each model consumes a portion of your monthly word balance based on how much processing it requires. The more advanced the model, the more words are used.
Formula
[Number of words generated] × [Model Multiplier] = Words used
This applies to both your message and the AI's response. The more you chat—and the longer your messages—the more words you'll use.
What Is the “Multiplier”?
Every AI model has a multiplier. It's our internal system for measuring how intensive the model is to run. More powerful models have a higher multiplier, which means they consume more of your balance per word.
We keep this system in place to give you flexibility:
Use faster, more affordable models for lightweight tasks
Switch to smarter, more powerful ones when needed
Model Usage Table
Below is a general estimate of how many words each model uses in a short 15-word interaction:
Model Name | Multiplier | Words Used per Short Exchange |
o1 pro | 50× | 750 words |
gpt-4.5 | 15× | 225 words |
o3 pro | 6.67× | 100 words |
claude 4 opus | 6× | 90 words |
o1 | 5× | 75 words |
gpt-4 turbo | 5× | 75 words |
claude 3 opus | 4× | 60 words |
claude 3 sonnet | 4× | 60 words |
llama 3 70b | 3.5× | 52.5 words |
claude 2.1 | 3× | 45 words |
gpt-4 | 3× | 45 words |
gemini 1.5 pro | 2.5× | 37.5 words |
mixtral | 2× | 30 words |
gemini 1.5 flash | 1.5× | 22.5 words |
command-r+ | 1.2× | 18 words |
gpt-4o (baseline) | 1× | 15 words |
gemini 1 pro | 0.9× | 13.5 words |
command-r | 0.7× | 10.5 words |
gpt-4.1 | 0.7× | 10.5 words |
zephyr | 0.5× | 7.5 words |
openchat | 0.5× | 7.5 words |
llama 3 8b | 0.4× | 6 words |
mistral | 0.4× | 6 words |
gemma | 0.3× | 4.5 words |
gpt-4.1 mini | 0.2× | 3 words |
gpt-3.5 turbo | 0.2× | 3 words |
gpt-4.1 nano | 0.1× | 1.5 words |
Note: Actual usage depends on message length. This table assumes a typical 15-word exchange.
Estimating Message Capacity
To estimate how many messages you can send:
Message Capacity = Word Balance ÷ (Message Length × Multiplier)
Example with 100,000 words:
GPT-4o (1×) →
100,000 ÷ 15 = 6,667 messages
GPT-4.5 (15×) →
100,000 ÷ 225 = 444 messages
GPT-4.1 nano (0.1×) →
100,000 ÷ 1.5 = 66,667 messages
A Note About Chat History
Every message you send and every message you receive adds to your word count.
The more context the model holds onto, the more it costs. This is especially important with high-multiplier models.
Long conversations = more words retained = higher word usage
Use compact models or start a new chat when starting a new topic
How to Make the Most of Your Balance
Use low-multiplier models for fast, simple tasks
Use powerful models only when necessary
Buy top-up word packs (10K, 25K, or 50K)
Track Your Usage
Go to Team Page → Usage to:
View your word balance
Add more words any time
Have any questions or concerns? Reach out to our team via chat —we’re happy to help. 💬✨.