> For the complete documentation index, see [llms.txt](https://docs.3techagency.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.3techagency.com/documentation/start-here/connect-ai-and-choose-models.md). # Connect AI and choose models Get an API key, connect a provider through WordPress Connectors and understand the two model tiers. WP AI Manager does not host AI. It uses the **WordPress core AI Client** and the AI Connector plugins you install. The connector holds your provider key; WP AI Manager never asks for it. Any provider with a WordPress AI Connector works, for example OpenAI, Anthropic, Google or OpenRouter. OpenRouter is the easiest way to try several providers behind one key. ## Connect a provider {% stepper %} {% step %} ### Get an API key from your provider Create an account with the provider you want to use, add billing if it asks, then create an API key. Copy the key when it is shown: most providers display it only once. | Provider | Create your key | Billing | | ---------------------------------------- | ------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------- | | **OpenAI** (GPT, gpt-image) | [platform.openai.com/api-keys](https://platform.openai.com/api-keys) | Prepaid credits. Add a payment method under **Settings > Billing** in the OpenAI platform before the key works. | | **Anthropic** (Claude) | [platform.claude.com/settings/keys](https://platform.claude.com/settings/keys) | Add a payment method or credits under **Settings > Billing** in the Claude Console. | | **Google** (Gemini, Imagen) | [aistudio.google.com/apikey](https://aistudio.google.com/apikey) | Free tier with lower rate limits. Turn on billing in Google AI Studio for higher limits. See [Gemini pricing](https://ai.google.dev/pricing). | | **OpenRouter** (many providers, one key) | [openrouter.ai/keys](https://openrouter.ai/keys) | Prepaid credits from your OpenRouter account. Manage keys at [openrouter.ai/settings/keys](https://openrouter.ai/settings/keys). | **Which one should I pick?** * **Just trying it out:** OpenRouter. One key reaches models from several providers. * **Image generation:** OpenAI or Google. Both offer photo models. * **Already pay a provider:** use that one. {% hint style="warning" %} An API key is like a password that spends your money. Never paste it into AI Chat, a prompt, Site Knowledge, a screenshot or a support ticket. Set a monthly spending limit in your provider account. If a key leaks, delete it at the provider and create a new one. {% endhint %} {% endstep %} {% step %} ### Install a connector Install and activate the WordPress AI Connector plugin for that provider. {% endstep %} {% step %} ### Add the key in WordPress Open **WordPress Settings > Connectors** (not WP AI Manager Settings), paste the API key or sign in, and save. {% endstep %} {% step %} ### Approve outbound access if asked If your site uses **Connector Approval**, approve the provider connector. The approval belongs to the connector plugin, not to WP AI Manager. {% endstep %} {% step %} ### Check the status in Chat Open **WP AI Manager > AI Chat**. With a working provider the status dot is green and the composer is active. Try "List my recent pages". {% endstep %} {% endstepper %} With no provider connected, Chat says **No AI provider connected** with a grey dot, and a **Connect an AI provider first** card with **Open Connectors** replaces the composer. ## The two model tiers You do not have to pick a model. Chat offers two modes, and each resolves to a recommended model for the first connected provider. | Tier | Used for | | ---------------------- | ---------------------------------------------------------------------------- | | **Best quality** | Chat, page builds and edits, planning, anything that needs care | | **Faster, lower cost** | Quick questions and per-item Task work such as meta descriptions or alt text | Recommended models when nothing is set in Settings (tried in order; a dated snapshot of the same model also matches): | Provider | Best quality | Faster, lower cost | | --------- | ----------------------------------------------------- | --------------------------------------------------------------------------- | | Anthropic | claude-sonnet-5-5, claude-sonnet-5, claude-sonnet-4-5 | claude-haiku-5-5, claude-haiku-5, claude-haiku-4-5 | | OpenAI | gpt-6.1-sol, gpt-6-sol, gpt-5.6-sol, gpt-5.4 | gpt-6-luna, gpt-5.6-luna, gpt-5.4-mini, gpt-5.4-nano | | Google | gemini-3.1-pro, gemini-3-pro, gemini-2.5-pro | gemini-3.7-flash, gemini-3.5-flash, gemini-3.5-flash-lite, gemini-2.5-flash | For another provider, Faster picks a model whose name signals a low-cost line (mini, nano, flash, lite, haiku, luna, small); Best quality uses the provider's default. ## Choose specific models (optional) Open **WP AI Manager > Settings > AI Models** and expand **Advanced: choose specific models**. Each select shows **Recommended: model** so you can see what is used today. Leave both fields empty to keep the recommendations. A model you choose always wins. The same tab lists **Connection details** per provider (Ready, Models, Problem), the **Image models** in the order they are tried, and the **Last image problem**. Use **Check again** after changing a connector. **Settings > Advanced > Model choice in chat** is off by default. Turn it on to let chat users pick any model. The list then hides dated snapshots and audio, image, embedding, realtime, codex and search models. {% hint style="warning" %} Choose models that support tool use. Image understanding (alt text, reading screenshots) needs a vision model; Settings marks those **(vision)**. Provider prices change independently of WP AI Manager. Set a budget in your provider account as well as the plugin's monthly token limit. {% endhint %} ## If the connection fails | Message or symptom | Cause and fix | | ---------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------- | | No AI provider connected | No connector installed or configured. Use **Open Connectors**. | | Connector present, requests denied | Connector Approval is pending or denied for the connector plugin. | | Key rejected | Invalid, expired or deleted key, or no billing on the provider account. Check billing, then replace the key in Settings > Connectors. | | Rate limited | The plugin retries once. Wait, or raise the limit with your provider (a free Gemini tier has low limits). | | Conversation too long | The plugin retries on a shorter history. Start a new chat for unrelated work. | | No provider models listed | Fine on its own: the recommended model is still used. Use **Check again**. | See [Troubleshooting](/documentation/reference-and-help/troubleshooting.md) for more. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.3techagency.com/documentation/start-here/connect-ai-and-choose-models.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.