ClassifAI 3.5.0 Adds Google AI Imagen Image Generation

ClassifAI 3.5.0 expands the plugin’s AI image generation options with Google AI Imagen as a new provider, giving WordPress teams another way to create images from prompts inside their publishing workflow. This release also improves generated titles and excerpts, adds more flexible audio transcription via WP-CLI, and includes several fixes that make Key Takeaways and API errors easier to understand.

Feature highlights

The biggest addition is Google AI Imagen support for the Image Generation feature. After updating, you can select Google AI Imagen as your provider, add a valid Google AI API key, and generate images from the standalone image generation page or supported editor workflows. A paid Google AI account is required for Imagen access.

• Google AI Imagen for image generation — Generate and import AI-created images using Google AI Imagen as a new Image Generation provider.
• Audio transcription from URLs or file paths — The WP-CLI transcription command can now process audio files from remote URLs or local file system paths.
• Clearer setup guidance — The Installed Plugins screen now prompts users to register ClassifAI so setup status is easier to notice.

Improved writing assistance

ClassifAI’s content-generation features also receive better defaults aimed at producing more polished editorial output with less prompt tweaking.

• Sentence case titles — Title generation prompts now default to sentence case formatting for more consistent headline style.
• Excerpts with author context — Excerpt Generation can now include the post author’s name as context when creating summaries.

Other improvements and fixes

This release includes a round of quality-of-life updates focused on clearer errors, safer processing, and better compatibility with provider limits.

• OpenAI image processing model update — OpenAI-powered image processing features now use gpt-4.1-mini instead of gpt-4o-mini.
• Better Key Takeaways behavior — The Key Takeaways block now shows a clearer error state and does not render on the front end when there are no takeaways to display.
• More readable API errors — API error messages are no longer over-escaped, making provider responses easier to interpret.
• Embedding generation safeguards — API errors during embedding generation now stop processing correctly instead of risking repeated loops.
• Cleaner settings labels — Post status labels shown in settings now have HTML stripped and entities decoded for clearer display.
• OpenAI image type validation — OpenAI image processing now limits uploads to supported JPG, PNG, GIF, and WebP files.

For developers

• Key Takeaways auto-run filters — New filters classifai_chatgpt_key_takeaways_auto_run, classifai_ollama_key_takeaways_auto_run, and classifai_azure_openai_key_takeaways_auto_run allow automatic Key Takeaways processing to be disabled.
• Corrected DALL·E hook name — The hook name has been corrected from classifai_openai_dalle_embeddings_model to classifai_openai_dalle_model.
• New developer documentation — Documentation now covers how to trigger ClassifAI features from your own UI and organizes project diagrams under a main diagrams page.

See the full release notes on GitHub.

Learn more and download ClassifAI.