Embedded Files
1. Install the Add-on
➡️ Install the AI Sheets Add-on from Google Workplace Marketplace
https://workspace.google.com/marketplace/app/ai_sheets/361369486855
To launch the Add-on
➡️ Open a Google SheetsTM document and select:
Extensions > AI Sheets > API Settings
2. Add your OpenAI Key
In the AI Settings sidebar:
➡️ Enter your API key/s
OpenAI API: API Keys can be purchased and generated: https://platform.openai.com/settings/organization/api-keys
Perplexity API (Optional): API Keys can be purchased generated: https://www.perplexity.ai/settings/api
➡️ Save Keys
Note: API keys are pay as you go > $5 to get started will give you plenty of credit to test all the features and get started on your project.
2. Start using Custom Functions
Select any cell and start typing "=" and the first letters of the function.
An auto-complete drop-down will show all functions available.
Guide to Custom Functions
Need help gettign started? Get in touch! hello@alicemrongovius.com
Model Comparisons: Cost & Performance
Model/API Pricing Tier Speed & Performance Best For
GPT-4 Turbo Premium Fast In-depth, complex tasks; premium blog posts and SEO content
GPT-4o Mid-range Balanced speed and performance. General content creation; reliable output without premium cost
GPT-4o-mini Budget Cost-effective, slightly reduced performance Routine tasks and basic content generation; when cost is a key concern
Perplexity API Varies* Depends on specific implementation Quick research, summaries, and simpler content tasks
Pricing:
OpenAI https://openai.com/api/pricing/
Perplexity https://docs.perplexity.ai/guides/pricing
Understanding Key Parameters: Max Tokens & Temperature
Max Tokens
The maximum number of “tokens” (roughly equivalent to pieces of words) that the model can output. Think of it as setting a cap on the length of the response.
Usage:
Lower max tokens (e.g., 100) give you shorter, concise outputs.
Higher max tokens (e.g., 300 or more) allow for more detailed responses.
Consideration:
Higher token limits can increase both the response time and cost, so choose a value that fits your needs.
Temperature
A parameter that controls the randomness of the model’s responses.
Usage:
Low Temperature (0.2 – 0.5):
Yields more deterministic and focused responses, good for factual summaries or tasks where consistency is key.
High Temperature (0.7 – 1.0):
Produces more creative and varied outputs, which can be useful for brainstorming or when you need a bit of unpredictability.
Consideration:
Adjusting temperature can help tailor the tone and creativity of the output to your specific use case.
Page updated
Google Sites
Report abuse