First Setup

SailFish does not include a built-in AI model — you need to configure an AI service API, like inserting a SIM card into a phone to make calls.

Don’t worry; the basic model setup usually only takes a few minutes. Web search can be configured right after that, or added later.

What is an API Key

An API Key is a secret string your AI provider gives you to verify your identity. Think of it as an “AI membership card” — with it, SailFish can call the AI service to think and answer.

Most AI services bill by usage (like mobile plans). Actual cost depends on how often you use it, so it is a good idea to monitor your balance or set budget alerts in the provider dashboard.

Recommended: DeepSeek

If this is your first time using an AI API, we recommend starting with DeepSeek:

• Simple registration, no VPN required for most regions
• Very low cost (around 1 CNY per million tokens)
• Good general capability for most everyday tasks

SailFish supports dozens of AI providers (OpenAI, Claude, Qwen, Gemini, etc.). You can switch or configure multiple at any time. We use DeepSeek as an example; the flow is similar for others.

Step 1: Register for DeepSeek and Get an API Key

1. Open DeepSeek Platform
2. Click Register in the top-right and complete registration (e.g., phone number)
3. After login, go to the console
4. In the left menu, click API Keys
5. Click Create API Key and give it a name (e.g., “SailFish”)
6. Copy the generated API Key — the string starting with sk- is your key

Keep your API Key secure and do not share it. If it is leaked, delete the old key in the DeepSeek console and create a new one.

7. New users may receive free credits. If balance is low, top up according to the provider console instructions

Step 2: Configure in SailFish

1. Open SailFish
2. Click the Settings icon (gear) at the bottom-left
3. Find the “AI Model Configuration” section in Settings
4. Click “Add”
5. Select “DeepSeek” in the provider list — the system will fill in the API URL and default model
6. In the API Key field, paste the API Key you copied
7. Choose model deepseek-v4-flash (usually selected by default)
8. Click “Save”

Step 3: Set as Active

After saving, you will see the new configuration in the model list. Click the “Use” button on the right (or the checkmark icon) to make it the active model.

Setup complete! SailFish can now “think.”

Recommended Add-on: Web Search (Google, Highly Recommended)

With a search engine configured, the Agent can fetch real-time information from the web instead of relying solely on its training data.

Get a Google Custom Search API Key

1. Go to Google Cloud Console and create a project
2. Enable the Custom Search API for the project
3. Create an API Key under Credentials and copy it
4. Go to Programmable Search Engine, create a new search engine set to search the entire web, and copy the Search Engine ID (cx)

Configure Google Search in SailFish

1. Open Settings → Web Search
2. Select “Google Custom Search” as the search engine
3. Paste your API Key and Search Engine ID
4. Save

Once configured, the AI will automatically call Google when it needs real-time information.

Verify the Model Configuration

Return to the main interface and try typing something, for example:

Hello, please introduce yourself

If the AI responds normally, the configuration is successful.

If there is no reply or an error:

• Check that the API Key is copied completely (starts with sk-)
• Check that your DeepSeek account has sufficient balance
• If you are on a corporate network, you may need to configure a proxy (see “Network Proxy” below)

Other Common Providers

Besides DeepSeek, here are other common AI providers you can configure in SailFish:

You can configure multiple models: use a cheaper one for daily tasks (e.g., DeepSeek) and switch to a stronger one (Claude, GPT-4o) for complex work. Switch in the top model selector in SailFish.

Local Deployment (Ollama)

If you want fully free, offline AI, you can run a model locally with Ollama:

1. Go to ollama.com and download/install Ollama
2. Run ollama run qwen2.5 in a terminal to download and start a model
3. In SailFish, add a new AI config and select the “Ollama” template
4. Leave API Key empty; set model to qwen2.5
5. Save and set as active

Local model quality depends on your hardware. At least 16GB RAM is recommended. A GPU improves performance noticeably.

Network Proxy

If you need a proxy to reach overseas AI services (e.g., OpenAI, Claude), you can set a proxy per AI config:

1. Edit the corresponding AI configuration
2. In the Proxy field, enter the proxy URL (e.g., http://127.0.0.1:7890)
3. Save

Proxy settings are per config — you can have DeepSeek use direct connection and OpenAI use the proxy.

Next Step

Configuration complete! Next, get to know the interface and learn the main areas of SailFish.