Setting Up Your API Key
Sieve uses AI to classify your images. To do this it needs access to an AI provider — a service that runs the vision model. You connect your own account with your own key, so Sieve never sees your images on its own servers.
This page explains how to get a key for each supported provider and enter it into Sieve.
Which provider should I choose?
OpenRouter
Free tier available
Getting started, no credit card needed
Ollama
Free, runs locally
Private or sensitive images, offline use
OpenAI
Paid
High accuracy, willing to pay
Anthropic
Paid
High accuracy, willing to pay
Google Gemini
Free tier with limits
Already have a Google account
If you are unsure, start with OpenRouter. It is free, takes two minutes to set up, and works well for most use cases.
If your images are sensitive — patient data, proprietary research, unpublished results — use Ollama. It runs entirely on your own machine and your images never leave your device at all.
OpenRouter (recommended for getting started)
OpenRouter gives you access to several powerful vision models for free without requiring a credit card.
Go to openrouter.ai and create a free account
Once logged in, go to Keys in the left sidebar
Click Create Key, give it a name like
sieve, and copy the keyOpen Sieve, go to Settings → API Key
Select OpenRouter as your provider
Paste your key and save
The default model in Sieve for OpenRouter is qwen/qwen2.5-vl-72b-instruct:free which is a strong free vision model. You can change the model in Settings if you want to try others.
Free tier limits: OpenRouter's free models allow around 20 requests per minute and 200 per day. For most users this is more than enough. If you are processing very large batches you may hit the daily limit — in that case either wait until the next day or switch to a paid model.
Ollama (local, offline, private)
Ollama runs an AI model directly on your computer. No internet connection is needed after setup, and your images never leave your device under any circumstances.
This is the right choice if:
You work with sensitive, confidential, or patient images
Your institution has data policies that prevent using cloud services
You want complete privacy and control
Setup:
Install Ollama from ollama.com
Open a terminal and pull the vision model:
Ollama runs automatically in the background once installed
Open Sieve, go to Settings → API Key
Select Ollama as your provider
No API key is needed — just save
Note on speed: Ollama runs on your own hardware, so classification will be slower than cloud providers, especially on older machines. On a modern laptop with a decent GPU, expect a few seconds per image. On older hardware it may take 10–20 seconds per image.
OpenAI
Go to platform.openai.com and create an account
Add a payment method and purchase some credits (even $5 will last a long time for image classification)
Go to API Keys and create a new key
Open Sieve, go to Settings → API Key
Select OpenAI as your provider
Paste your key and save
Sieve uses gpt-4o for OpenAI which has strong vision capabilities.
Anthropic
Go to console.anthropic.com and create an account
Add a payment method and purchase credits
Go to API Keys and create a new key
Open Sieve, go to Settings → API Key
Select Anthropic as your provider
Paste your key and save
Sieve uses claude-sonnet-4-6 for Anthropic.
Google Gemini
Go to aistudio.google.com and sign in with your Google account
Click Get API Key and create a new key
Open Sieve, go to Settings → API Key
Select Gemini as your provider
Paste your key and save
Gemini has a free tier but with rate limits. If you hit them frequently, consider switching to OpenRouter or Ollama.
Where is my key stored?
Your API key is stored locally on your device using secure encrypted storage. It is never sent to any Sieve server because Sieve has no servers. The key is only used to make direct API calls from your device to your chosen provider.
Changing your provider or key
Go to Settings → API Key at any time to change your provider or update your key.
Next: How Classification Works
Last updated