AI Settings
Configure OpenAI, Gemini or Claude through the Framework for HuntBee AI-powered features.
- 1Choose a supported provider—OpenAI, Gemini or Claude—and enter its API key in Framework → AI Settings.
- 2Select the model you have access to, save and click Test Connection.
- 3Open the SEO Suite AI Content Generator, switch to AI mode, and preview a small request.
The connection test succeeds and a small preview works. A saved key alone is not proof that the provider is responding.
Tip: Templates mode in SEO Suite works without any provider or API credits. AI calls can incur provider charges.
Configure one supported AI provider here. SEO Suite AI Content uses this shared Framework connection.
Start with one provider
You do not need to configure every AI provider. Choose the service you already use, enter its API key, enable it and test the connection.
Supported model choices in Framework 4.2.2
Current options include GPT-6.1 Sol (OpenAI), Gemini 3.8 Flash (Google) and Claude Sonnet 5.5 (Anthropic), alongside other models listed in your installed version. The model you can actually use depends on your own API account and provider access.
Shared AI settings
| Setting | What it means | Recommended starting point | Change it when… |
|---|---|---|---|
| Default Provider | Which enabled provider HuntBee uses when a Suite does not explicitly request another one. | Choose the one provider you intend to use most. | You want a different default provider. |
| Provider Status | Turns that provider connection on or off. | Enable only providers you have configured and tested. | You stop using a provider or temporarily want to prevent calls. |
| API Key | Secret credential issued by the AI provider. It is stored in OpenCart settings and should never be public. | Enter the provider key, save, then test. | The provider rotates/revokes your key. |
| Model | The AI model used for requests. The Framework lists currently supported/recommended choices for your installed version. | Use the current recommended general model shown in the Framework unless you have a reason to choose another. | You need a different speed/cost/quality balance. |
| Reasoning | Where the selected model supports it, controls how much internal reasoning effort is requested. | Medium is a balanced starting point. | Your chosen model supports it and a task benefits from a different effort. |
| Verbosity | Where supported, controls how concise or detailed model output should be. | Medium. | Your content workflow consistently needs shorter/longer output. |
| Max Output Tokens | Optional cap on generated output. Blank/0 lets the provider/model default apply; Claude uses a safe fallback when needed. | Leave blank/0 at first. | You specifically need to limit response size/cost. |
| Temperature | Optional creativity/randomness control, only for compatible models. | Leave blank/default unless you understand your model behaviour. | A compatible model and workflow needs more/less variation. |
| AI Logging | Records safe diagnostic information without logging secrets. | Keep enabled while setting up or troubleshooting. | You have a reason to reduce diagnostic logging after stable use. |
Test Connection
The test sends a very small request using the currently entered configuration. A successful test confirms that HuntBee can authenticate and receive a response. It does not guarantee that every future prompt will succeed, because provider limits, model availability and billing can change.
Cost and privacy
- AI provider usage can create charges on your provider account. HuntBee does not control the provider’s pricing.
- Only send catalog/business data you are comfortable processing through the provider you selected.
- Never paste API keys into public screenshots or support forums.
- Use AI Content preview/manual generation before enabling automatic queueing.