To use ThreadKeeper's AI summary feature,
you'll need an API key from one of the supported AI providers.
Choose your provider and follow the instructions below.
Offers a free tier with fast response times.
The most recommended provider for first-time users.
Recommended model: Gemini 2.0 Flash Lite
Go to https://aistudio.google.com/
and sign in with your Google account.
The API key may only be displayed once.
Be sure to copy it and save it in a safe place.
Open ThreadKeeper's settings screen,
select "Google Gemini" under "AI Provider,"
and paste your API key to save.
// Example config (settings.json)
{
"ai_provider": "gemini",
"ai_model": "gemini-2.0-flash-lite",
"api_key": "AIza..."
}About the free tier: The Gemini API has a limit of 15 requests per minute
and a daily token cap.
Normal usage with ThreadKeeper easily stays within the free tier.
Supports GPT-4o / GPT-4o mini.
Ideal when high-accuracy summaries are needed.
API usage incurs pay-as-you-go charges.
Go to https://platform.openai.com/
and create an account or sign in.
The API key is shown only once when created.
If lost, you will need to create a new one.
Go to Settings, then Billing to add a payment method.
Credit or debit cards are accepted.
In ThreadKeeper's settings, select "OpenAI"
and paste your API key to save.
// Example config (settings.json)
{
"ai_provider": "openai",
"ai_model": "gpt-4o-mini",
"api_key": "sk-..."
}Cost estimate: With GPT-4o mini,
each summary costs approximately $0.0001 to $0.0003.
Even with 10 summaries per day, the monthly cost is well under $0.10.
Supports Claude 3.5 Sonnet / 3.5 Haiku.
Known for natural and polished summaries.
API usage incurs pay-as-you-go charges.
Go to https://console.anthropic.com/
and create an account or sign in.
Go to Settings, then Billing to purchase credits.
The minimum purchase is $5.
In ThreadKeeper's settings, select "Anthropic"
and paste your API key to save.
// Example config (settings.json)
{
"ai_provider": "anthropic",
"ai_model": "claude-3-5-haiku-latest",
"api_key": "sk-ant-..."
}Cost estimate: With Claude 3.5 Haiku,
each summary costs approximately $0.0002 to $0.0005.
Runs completely offline and entirely free.
No data is ever sent externally.
Requires a reasonably powerful PC.
Download and install the Windows version from
https://ollama.com/.
Open Command Prompt or PowerShell
and run the following commands:
# Recommended: lightweight and fast
ollama pull gemma3:4b
# For higher quality summaries
ollama pull llama3.2
# Model with strong multilingual support
ollama pull qwen2.5:7bSystem requirements: For the 4B model,
8GB of RAM or more is recommended.
For the 7B model, 16GB or more is recommended for comfortable usage.
Once installed, Ollama starts automatically
in the background.
If you see the Ollama icon in the taskbar, you're good to go.
In ThreadKeeper's settings, select "Ollama."
No API key is needed. Just set the model name and you're done.
// Example config (settings.json)
{
"ai_provider": "ollama",
"ai_model": "gemma3:4b",
"ollama_url": "http://localhost:11434"
}Downloading a model with Ollama may take
several minutes to tens of minutes.
This is only for the first time; afterwards, it's available offline instantly.
API Key Management
API keys are stored locally in ThreadKeeper's settings file.
They are never sent to ThreadKeeper's servers.
Do not share your API keys with others.
Data Sent to AI
Only the following data is sent to the AI:
titles of open applications,
browser tab titles and URLs,
and the first 200 characters of the clipboard.
File contents and passwords are never sent.
When using Ollama, no data leaves your PC at all.
Switching Providers
You can change your AI provider at any time from the settings.
Session data is independent of the provider,
so your existing data remains intact when switching.