What's New in AI
Create and collaborate on presentations in real time
Turn a conversation, outline, or files into an editable presentation, then refine it with teammates using shared comments and live collaboration. Present directly or export to PowerPoint, PDF, or Google Slides.
GPT-6 Pro gets faster answers and parallel agents
The Pro model now responds more quickly, uses plugins and tools more effectively, and can split complex work across multiple agents running in parallel. You can follow each agent’s progress and inspect its completed work.
Cloud agents can build interactive canvases and use larger environments
Cloud agents can now create shareable interactive canvases, while a new beta API lets teams programmatically manage their environments. Environments also support up to 300 repositories and load large setups five times faster.
Add voice assistants, live web research, video, and music to your apps
New AI models and tools let you build apps with real-time voice conversations, cited web answers, video generation, original music, and smarter research or support workflows. Users can choose a model or describe the capability they want in chat.
Haiku 4.5 leaves the model picker on October 29
Haiku 4.5 will be removed from the model selector in chat, Cowork, and Design for Free, Pro, and Max users. Existing chats remain available, but continuing them after October 29 will use Haiku 5.5 instead.
GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini retired on October 15
These older models are no longer available in ChatGPT, so users must switch to newer models for existing and new conversations. The change does not affect the API.
Voice gets GPT-Live-1.1, more natural accents, and 13 new voices
Voice conversations now sound more natural across languages, with improved accents in French, German, Hindi, Japanese, Korean, Mexican Spanish, and more. New voice choices include Bubbie in English, and Live gives shorter updates when it is searching.
Grok Voice 3.0 brings faster, more natural speech conversations
Grok Voice 3.0 is now available for speech-to-speech interactions, with sub-second latency, improved turn-taking, and stronger performance on voice tasks. The API price stays the same, and the grok-voice-latest alias now uses the new model automatically.
U.S. voters will get official election information in chat later this month
Claude will begin showing state-specific voting details from Democracy Works, including registration, polling-place, and ID information, directly in U.S. conversations. It will also provide current official election results on election night and afterward.
Create growth plans and marketing campaigns from your app
Lovable can now help attract customers by creating growth plans and producing search, video, advertising, landing-page, and social content. It can also connect to channels such as Google, Facebook, and Instagram to help carry out the plan.
There's more happening in AI
Create a free account to browse the full archive, filter by date, and get updates the moment they drop.
Get full access