AI MODELS FOR WORDPRESS
Rewrite, summarize or translate imported RSS content with OpenAI, Claude, Gemini, Grok, OpenRouter or your own OpenAI-compatible endpoint. Choose the model for each step in your WordPress publishing workflow.
Use your own accounts and model IDs. Pay providers directly.
A Hugging Face model served behind an OpenAI-compatible chat API. An LM Studio server reachable from WordPress. Set the endpoint URL, credentials and model — then use it in your prompts.
How custom endpoints workEvery service runs on your own account: OpenRouter, OpenAI, Anthropic, Google Gemini, xAI, Stability AI, GoAPI and CometAPI for AI; DeepL, Google Translate and Yandex Translate for translation; WordAi, SpinRewriter and SpinnerChief for spinning; Pixabay for stock photos. You pay each provider at its own rates. No markup, no credits to buy from us.
Give every service a request limit per hour, per day or per month. When a limit is reached, RSS Retriever stops calling that service until the next period starts, so a busy feed or a long multi-section article cannot drain your balance overnight. Custom endpoints take an ID, a URL, a default model and a key, and work in every AI shortcode.
No key, no AI: feed import, full-text extraction, filters, templates, Spintax and the built-in synonymizer work without any external service.
Accounts & request limits
Click to enlargePROMPT PIPELINES
Give [gpt_article] a topic from your feed and your editorial instructions. RSS Retriever generates an outline, then writes each section in a separate AI request, passing the text already written into the next step.
The first request generates section headings from the topic, section count and your instructions. The structure parameter guides the outline; directives supplies shared writing rules.
[gpt_article
topic="%post_title%"
sections="5"
structure="Open with the problem;
end with practical next steps"
directives="Write clear HTML paragraphs.
Use this source as factual context:
%post_content_notags%"]Each writing request includes the topic, full outline, current section, instructions and accumulated article. This gives the model context for transitions and helps it avoid repeating earlier points.
As the article grows, later requests carry more input text. Length remains subject to the chosen model's context limits.
01 Write section 1
Topic + outline + instructions
Current section: 1
02 Write section 2
Topic + outline + instructions
Current section: 2 + text of section 1
03 Write section 3
Topic + outline + instructions
Current section: 3 + text of sections 1-2
Continue through the outline.Use section1, section2 and other numbered parameters to add instructions to the shared directives. Ask one section for a comparison table and another for an actionable checklist.
[gpt_article
topic="%post_title%"
sections="5"
directives="Use clear HTML. Preserve facts
from this source: %post_content_notags%"
section1="Explain the problem and who it affects"
section2="Compare the options in an HTML table"
section5="Finish with an actionable checklist"]Every part of the post has its own template: title, slug, content, excerpt, SEO title and meta description. Combine your HTML with more than 30 placeholders — feed fields, XML tags and attributes, custom fields, feed URL parameters, dates in any PHP format, text shortened to N words — and AI shortcodes for OpenAI, Claude, Gemini, Grok, OpenRouter and your own endpoints.
Drag title, content and excerpt into the order you need: write the article first, then let the title and the excerpt summarize the finished text. Feed models plain text with %post_content_notags% and shortened input to save tokens. Spintax varies the wording from post to post. If an AI request fails, the item is not published half-finished; it is tried again on the next run.
Click to enlargeAI IMAGE GENERATION FOR WORDPRESS POSTS
Import a plain-text article and turn it into an illustrated post. With batch insertion and %gpt_image_prompt%, RSS Retriever builds a separate image prompt from the text of each section.
Earlier prompts are considered to reduce repetition. You choose the insertion tag, image count and visual style.
Generate with Midjourney (Relax, Fast or Turbo mode) and FLUX through GoAPI or CometAPI, OpenAI GPT Image and DALL·E, Stability AI (Stable Image Ultra, Core, SD 3.5, SDXL) or any image model on OpenRouter: Gemini image models, Seedream, FLUX Pro and more.
Put one image above or below the article, replace the first or every image of the source, or batch-insert a picture before every H2. Set the aspect ratio, resolution, negative prompt and style preset where the model supports them, add your own CSS class, or save the image URL to the thumb field to make it the featured image. Generated files are stored in your uploads; PNG images can be converted to WebP or JPEG.
Click to enlargeSearch YouTube with the post title, an AI key phrase (%gpt_keyphrase%) or any query you write, and embed the video above or below the article — or use its preview image as the featured image. A relevant video keeps readers on the page longer and turns a short news item into a fuller post.
Mix sources in one post: free full-size photos from Pixabay (keyword, language, type, orientation, minimum size), Creative Commons images from Google Images and AI-generated pictures. Set a fallback image source and make the featured image mandatory, so no post goes live without a picture.
YouTube & media searchCONTENT RETRIEVAL
Import full articles into WordPress instead of just RSS excerpts. Extract the source text, clean its HTML, configure connections for difficult sources and retrieve older posts from WordPress archives.
Most feeds carry two sentences and a link. RSS Retriever opens the link and pulls the whole article with the Full-Text RSS script it can install for you, or with your own container rule. If the full text cannot be retrieved, the item is skipped, not published as a teaser.
Then clean the result: strip tags, remove blocks such as share boxes, balance broken markup and keep or remove links.
Route requests through your proxy list, send a browser or crawler user agent, a referrer and your own HTTP headers. When a site blocks bots or rate-limits your IP, the Wayback Machine fallback fetches the latest archived copy of the feed, the article or its images.
When a WordPress blog’s feed has too few new items, RSS Retriever reads its older feed pages and imports posts that already left the main feed. Import RSS, Atom, YouTube channels, any XML structure and compressed feeds; fix broken encodings on the fly.
Click to enlargeConnection profiles and Wayback fallback give you more retrieval options. Results depend on source access and the availability of an archived copy.
WORDPRESS POST TEMPLATES & CUSTOM FIELDS
Build the HTML from reusable parts. Map source data to custom fields. Choose the post type, taxonomies and a custom template offered by your theme.
<article class="product-story">
<h2>%xml_tags[product_name]%</h2>
<p class="price">%xml_tags[price]%</p>
[ai_generate prompt="Explain these specs: %post_content%"]
<a href="%link%">View the source</a>
</article>Use templates, placeholders and AI to build SEO titles and descriptions. RSS Retriever saves them in the fields of the detected SEO plugin.
CONNECTED TRANSLATIONS
Add a copy of the feed for each language. Translate with AI, DeepL, Google or Yandex; RSS Retriever links language versions from the same source item in WPML or Polylang.
AUTOMATE THE WORK. KEEP THE DECISIONS.
Require any or all of your keywords, or block items with stop words: matched as whole words in the title, content and excerpt, or as any part of the link. Accept or reject items by their feed tags, skip items older or newer than N days, shorter or longer than N characters. Duplicates are detected by link, by title or both.
Filters run before translation and AI, so rejected items never cost you an API request.
Click to enlargePreview the feed item as content, as the full extracted article, its attachments or the raw XML before you save. Choose the post type (WooCommerce products included), categories, tags, a fixed or random author, the status — publish, pending, draft or private — and comments.
Keep the feed date or use the import time, spread dates with a random offset, schedule posts into the future, and delete them automatically after a set number of hours: ideal for offers, events and job listings.
Click to enlargeUse the built-in synonymizer, Spintax, supported spinner services or the AI Spinner. Mix the tools your workflow needs, without forcing every import through an AI call.
Rewriting tools →Every feed has its own update interval and its own number of posts per update; the list shows when each one ran and when it runs next. Pull selected feeds on demand and watch the report, mass-modify any setting across selected feeds (only the options you tick change) and shuffle update times so feeds do not all run at once.
Export feeds with all their settings and import them on another site. Run updates with WordPress cron or a real server cron job, and open the Syndicator log of the last run whenever something needs a closer look.
RSS feed management guide
Click to enlargeOWN THE TOOLKIT
Every feature included. Free updates and dedicated support. Choose the number of sites; pay once for the plugin.
Your AI and translation providers bill separately. Licenses can cover your own sites or your clients’ sites.
The free edition imports RSS and Atom feeds into WordPress posts. Install it from the official plugin directory, try it on your site and switch to the full toolkit when you need AI rewriting, AI images and everything else on this page.
Download free Lite7-day money-back guarantee under our refund policy.