Documentation

AI models & API keys

Choose from a dynamic provider catalog instead of a fixed list bundled with the plugin. Discover new text models, compare them with your own prompts and choose the right model for each campaign.

A model catalog that stays current

AI Autoblogger retrieves model lists directly from OpenAI, Anthropic, Google, xAI and OpenRouter. New models compatible with the plugin's text-generation connections become available when the catalog refreshes. You do not have to wait for a plugin release to add each model name.

The selector groups models by provider; OpenRouter models are also grouped by vendor. Search by model name or ID to find the exact version you want. This selector is available for article generation and the article-title generator.

Choose a text model here for writing articles and generating titles. Save your provider key in Settings before selecting a model. Image models are selected separately in the Media tab.

Match your key to the connection

SelectionSettings field
OpenRouter text models and OpenRouter imagesOpenRouter API Key
Direct OpenAI text modelsOpenAI API key
Direct Anthropic modelsAnthropic Claude API Key
Direct Google modelsGoogle Gemini API Key
Direct xAI modelsxAI API Key
Midjourney or Flux through CometAPICometAPI API Key
Midjourney or Flux through GoAPIGoAPI API Key
FreepikFreepik API Key
Stable DiffusionStability AI API key

Save keys in Settings, then choose the provider and model in your campaign. A key from one service does not authorize requests to another. The same model accessed directly or through OpenRouter can have different pricing, access requirements and limits.

WordPress / AI Autoblogger
AI Autoblogger - settings Click to enlarge

Find and refresh models

  1. Save the API keys for the providers you want to use.
  2. Open a campaign's Post content tab and expand the AI model selector.
  3. Click Refresh list to retrieve the current catalog, including newly available models.
  4. Search by name or ID, select the model and check Max tokens.
  5. Save the campaign and generate a small draft to check quality and provider access.

The model list updates automatically. Click Refresh list to look for newly released models immediately or after adding a provider key.

Existing campaigns and retired models

Refreshing the list keeps your saved campaign selections. A model marked (legacy) is an older selection that is no longer in the current catalog. If it stops working, choose another model and save the campaign.

If the selector asks you to choose another model, select an available model before saving.

Max tokens and article length

Max tokens sets the maximum amount of text the AI can generate for each section, not for the whole article. Tokens are units of text, not words. A higher value gives a section more room for detail; a lower value suits shorter explanations. This is an upper limit, not a target length.

Use Number of text sections to control how many parts the article contains, and Max tokens to give each part enough space. Describe the desired depth in your prompt. For a detailed guide, allow more space per section; for a brief overview, use fewer sections and a smaller limit. Longer output generally takes more time and costs more to generate.

Image models have their own selector

Choose image models in Media. OpenRouter offers a dynamically updated image-model list. See the media guide for image prompts, style and placement.

Enable Live search to let the AI search the web for up-to-date information while writing. Use it for recent news, trends and subjects where current information matters. This option is available when you select an OpenRouter text model.

Leave it off when your article does not need current web information. Web searches may add to your OpenRouter usage charges.

Choose the model for the job

Test the same title, outline and prompts with a few models. Compare instruction following, language quality, useful detail, response time and API cost. A short news post and a substantial technical guide need different budgets; assign models per campaign instead of forcing every article through one default.

Provider moderation, availability and prices remain controlled by the external services. Review generated content before publishing, especially factual claims, quotations and references.