> ## Documentation Index
> Fetch the complete documentation index at: https://docs.notifuse.com/llms.txt
> Use this file to discover all available pages before exploring further.
# OpenAI
> Configure OpenAI to power blog content creation and email generation in Notifuse.
The OpenAI integration enables AI-powered features across Notifuse using GPT models. This integration powers the [AI assistant](/features/ai-assistant) for blog content creation, email design and web analytics.
## Configuration
Navigate to **Settings** → **Integrations** → **Add Integration** → **OpenAI**.
### Form Fields
| Field | Description | Example |
| - | - | - |
| **API Key** | Your OpenAI API key | `sk-proj-...` |
| **Model** | The model ID to use — free text, so any current or future model works | `gpt-4.1` |
| **Base URL** | Optional. The API endpoint, for OpenAI-compatible providers. Leave empty for OpenAI itself | `https://openrouter.ai/api/v1` |
| **Reasoning effort** | Optional. How much reasoning models think: None, Minimal, Low, Medium, High or Extra high | `Default` |
### Choosing a model
The Model field is free text rather than a fixed list, so you are never blocked from a model that shipped after your Notifuse version — type the ID exactly as your provider spells it.
| Model | Good for |
| - | - |
| **GPT-4.1** | The default. Reliable for MJML templates and long-form posts |
| **GPT-5.4** | Complex reasoning and the most demanding generations |
| **o4-mini** | Cheaper reasoning for shorter, simpler tasks |
See [OpenAI's pricing page](https://platform.openai.com/docs/pricing) for current rates.
Leave **Reasoning effort** on Default for non-reasoning models such as GPT-4.1. Providers that do not support the setting ignore it. If generations come back cut off, lowering the effort gives the visible answer more of the output budget.
The assistant panel shows a running cost for Anthropic and Gemini conversations, but not for OpenAI: with any model ID accepted — including every model on OpenRouter and any self-hosted runtime — Notifuse has no price list to compute it from. Track spend in your provider's dashboard.
## Features Enabled
Once configured, the OpenAI integration enables the following AI-powered features:
### Blog AI Agent
The AI agent can assist with blog content creation:
| Feature | Description |
| - | - |
| **Write blog posts** | Generate complete blog articles based on topics or brief descriptions |
| **SEO optimization** | Automatically generate SEO-friendly meta titles, descriptions, and keywords |
| **Content editing** | Improve, expand, or summarize existing content |
### Email Editor AI
Generate emails automatically using AI:
| Feature | Description |
| - | - |
| **Email generation** | Create email content from prompts or descriptions |
| **Subject lines** | Generate compelling subject lines for better open rates |
| **Content variations** | Create multiple versions for A/B testing |
### Web Analytics Assistant
Answer questions about your website traffic with the [Web Analytics assistant](/web-analytics/ai-assistant):
| Feature | Description |
| - | - |
| **Ask about traffic** | Query sessions, pages, sources and goals in plain language |
| **Period summaries** | A written analysis of what moved, against the comparison period |
| **Drive the dashboard** | Change the period, apply filters, or open an Explore report |
## Enhanced with Firecrawl
The OpenAI AI agent can be enhanced by adding the [Firecrawl integration](/integrations/firecrawl). This gives the AI agent web access capabilities:
| Capability | Description |
| - | - |
| **Web research** | Search the web for current information and trends |
| **URL scraping** | Extract content from specific web pages for reference |
| **Competitive analysis** | Research competitor content and strategies |
| **Fact verification** | Verify claims against current web sources |
When Firecrawl is configured, the following tools become available to the AI agent:
| Tool | Description |
| - | - |
| `scrape_url` | Scrapes a URL and returns its content as markdown |
| `search_web` | Searches the web and returns relevant URLs |
## Using OpenRouter
[OpenRouter](https://openrouter.ai) is a unified API gateway that gives you access to hundreds of AI models (Google Gemini, Meta Llama, Mistral, DeepSeek, and more) through a single OpenAI-compatible endpoint.
Since Notifuse's OpenAI integration uses the standard OpenAI SDK, you can point it to OpenRouter by setting the **Base URL** field:
### Form Fields for OpenRouter
| Field | Description | Value |
| - | - | - |
| **API Key** | Your OpenRouter API key | `sk-or-v1-...` |
| **Model** | Any model available on OpenRouter | `google/gemini-3-pro-preview` |
| **Base URL** | OpenRouter API base URL | `https://openrouter.ai/api/v1` |
This allows you to use any model supported by OpenRouter — including Google Gemini, Meta Llama, Mistral, DeepSeek, Qwen, and many others — directly from Notifuse's AI features.
The same field covers any other OpenAI-compatible endpoint: Azure OpenAI, or a self-hosted runtime such as Ollama, vLLM or LiteLLM.
Browse available models at [openrouter.ai/models](https://openrouter.ai/models) to find the best fit for your use case and budget.
The AI agent generates structured MJML email templates and full blog posts with HTML formatting. Only models with strong coding capabilities (e.g. GPT-4.1, Claude Sonnet, Gemini Pro) are smart enough to produce correct output. Smaller or weaker models may generate broken templates or malformed content.
## Prerequisites
Before configuring OpenAI:
1. **OpenAI Account**: Create an account at [platform.openai.com](https://platform.openai.com)
2. **API Key**: Generate an API key from the OpenAI dashboard
3. **Credits**: Ensure your account has sufficient API credits
## Usage Notes
API calls to GPT models count toward your OpenAI account usage. Monitor your usage in the OpenAI dashboard to manage costs.
For most use cases, **GPT-4.1** offers a good balance of capability and cost. Step up to a reasoning model for complex creative work, and down to a mini or nano variant for simple, high-volume operations.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.