Skip to main content

LLM Settings

Configure which AI model powers your LENS insights. By default, Sealmetrics uses a self-hosted DeepSeek model, but you can configure your own provider for potentially faster responses or specific model preferences.

Accessing LLM Settings​

  1. Click the gear icon to open Settings
  2. Click LLM Settings in the sidebar

Default Configuration​

By default, Sealmetrics uses DeepSeek Local via Ollama:

AspectDetails
ProviderSelf-hosted DeepSeek (via Ollama)
CostFree (included in your plan)
PrivacyMaximum - data never leaves Sealmetrics infrastructure
API KeyNot required

This default configuration provides:

  • Zero API costs
  • Full data privacy (no external API calls)
  • Reliable availability

Available Providers​

ProviderTypeAPI Key RequiredBest For
DeepSeek LocalSelf-hostedNoDefault, privacy-focused
DeepSeek CloudCloudYesFaster responses
Anthropic (Claude)CloudYesComplex analysis, nuanced insights
OpenAI (GPT)CloudYesGeneral-purpose, fast
Google GeminiCloudYesMulti-modal capabilities

Configuring a Custom Provider​

Step 1: Click Configure Custom Provider​

If you're using the default, click the Configure Custom Provider button.

Step 2: Select Provider​

Choose your preferred provider from the dropdown. Each provider shows:

  • Name and type (Self-hosted or Cloud)
  • Whether an API key is required
  • Link to obtain an API key

Step 3: Enter API Key (if required)​

For cloud providers, enter your API key:

  1. Click the link to get an API key from the provider's website
  2. Copy your API key
  3. Paste it in the API Key field

Your API key is encrypted and stored securely.

Step 4: Configure Model Overrides (Optional)​

You can override the default models for different insight complexity levels:

LevelDefault UseExample Override
SimpleQuick summaries, basic alertsgpt-4o-mini
ComplexTrend analysis, recommendationsgpt-4o
CriticalExecutive summaries, strategic insightsclaude-3-opus

Leave blank to use the provider's default models.

Step 5: Test Connection​

Before saving, click Test Connection to verify:

  • API key is valid
  • Provider is reachable
  • Response latency is acceptable

A successful test shows:

  • "Connection Successful" message
  • Response latency in milliseconds

Step 6: Save Settings​

Click Save Settings to apply your configuration.

Managing Your Configuration​

View Current Settings​

Your configuration card shows:

  • Provider name and type
  • API key status (configured or not required)
  • Enabled/disabled status
  • Model overrides (if any)
  • Custom base URL (if configured)
  • Last updated timestamp

Edit Settings​

Click Edit to modify:

  • Switch providers
  • Update API key
  • Change model overrides
  • Toggle enabled status

Remove Configuration​

Click Remove to delete your custom configuration and revert to the default self-hosted DeepSeek.

Enable/Disable​

Use the Enable Custom Configuration toggle to:

  • Enabled: Use your custom provider
  • Disabled: Fall back to default self-hosted DeepSeek

This lets you keep your configuration saved while temporarily using the default.

Advanced: Base URL Override​

For DeepSeek providers, you can specify a custom base URL:

ProviderDefault URLWhen to Override
DeepSeek Localhttp://ollama:11434/v1Custom Ollama installation
DeepSeek Cloudhttps://api.deepseek.com/v1Proxy or custom endpoint

Multi-Account Configuration​

If you manage multiple accounts:

  1. Select the account from the dropdown at the top
  2. Configure LLM settings for that specific account
  3. Each account can have its own provider configuration

Security​

API Key Storage​

  • API keys are encrypted using AES-256-GCM before storage
  • Only the last 4 characters are displayed (****XXXX)
  • Keys are never logged or exposed in API responses

Data Privacy​

ProviderData SentPrivacy Level
DeepSeek LocalNone (self-hosted)Maximum
Cloud ProvidersAnalytics summaries for insight generationStandard

Cloud providers receive aggregated analytics data (not raw events) to generate insights. No personally identifiable information is ever sent.

Troubleshooting​

"Connection Failed" Error​

  1. Verify your API key is correct and active
  2. Check the provider's status page for outages
  3. Ensure your network allows outbound API calls
  4. Try testing with a different provider

Slow Response Times​

If insights take too long:

  1. Check the test connection latency
  2. Consider switching to a provider with lower latency
  3. Cloud providers are typically faster than self-hosted

"API Key Required" Error​

Cloud providers require an API key. Get one from: