ChatGPT Projects just received its most significant update since launch, and the implications for productivity are substantial. OpenAI upgraded the Project feature, adding several important tools that should improve your productivity while using the chatbot. For those using Projects to organize research, manage code repositories, or coordinate complex creative work, these six new features fundamentally change what’s possible within the platform.
6 New ChatGPT Projects Features
1. Voice Mode Arrives in Projects
ChatGPT Advanced Voice Mode is now available in Projects. You can discuss your files and past chats with the AI by voice. This isn’t merely a convenience feature—it changes mobile workflows. Picture reviewing quarterly reports while walking, brainstorming product features during your commute, or dictating code documentation hands-free.
The implementation surpasses basic transcription. Voice mode maintains full project context, meaning you can reference specific documents, previous conversations, or custom instructions naturally. This feature is helpful for brainstorming, reviewing documents or asking questions on the go. The AI responds as if it’s been part of every conversation in your project, because effectively, it has been.
Unite AI/Alex McFarland
2. Enhanced Memory That Actually Remembers
The memory upgrade might be the most transformative addition. If you’re a Plus or Pro user, memory can now reference past chats in your project to keep responses focused and informed by your previous conversations. This creates genuine continuity across sessions.
Consider a marketing project spanning multiple campaigns. Previously, you’d need to remind ChatGPT about brand voice decisions, rejected concepts, or strategic pivots in each new chat. Now, the AI maintains this institutional knowledge automatically. ChatGPT now references your recent conversations to deliver responses that feel more relevant and tailored to you, extending this capability specifically within project boundaries.
3. Full Mobile Functionality
You can now upload files and switch models directly from the ChatGPT mobile app. This eliminates the desktop dependency that previously hampered field work. Architects can photograph site conditions and immediately integrate them into design projects. Journalists can upload interview transcripts on location. Researchers can capture and analyze data without returning to their desks.
The model-switching capability on mobile is particularly noteworthy. Users can toggle between GPT-4o for complex reasoning and faster models for quick queries, optimizing for either depth or speed based on immediate needs—all from their phones. This flexibility aligns with the best large language models available in 2025, where different models excel at different tasks.
Unite AI/Alex McFarland
4. Surgical Sharing Controls
Projects now let you generate a unique link to share a single conversation, without exposing the rest of your project’s files or instructions. This granular control solves a persistent challenge in collaborative AI work.
Consultants can share specific analyses with clients without revealing proprietary methodologies. Educators can distribute individual lessons while protecting complete curricula. Development teams can showcase features without exposing entire codebases. The sharing system respects both transparency needs and intellectual property concerns, making it ideal for AI collaboration tools in professional settings.
Unite AI/Alex McFarland
5. Expanded File Capacity and Intelligence
You can upload up to 20 documents per Project, but the real advancement lies in how ChatGPT processes these files. The system now better understands relationships between documents, cross-references information automatically, and maintains awareness of file hierarchies.
When you upload a file to a Project, the file will only apply to the context of that Project. This isolation prevents data leakage between projects while enabling deep integration within them. Financial analysts can upload years of reports, knowing ChatGPT will synthesize insights without contaminating other client work. This capability rivals dedicated AI tools for data analysts, but within the familiar ChatGPT interface.
6. Project-Level Custom Instructions
Instructions set in your Project will not interact with any conversations outside of your Project, and will supersede custom instructions set in your ChatGPT account. This hierarchical system enables unprecedented specialization.
A technical writer might configure one project for API documentation with specific formatting requirements, while another project focuses on user guides with entirely different voice and structure needs. Each project becomes a specialized AI assistant tuned for particular tasks.
Privacy Controls Address Enterprise Concerns
OpenAI won’t use information in Projects to improve ChatGPT by default for Team, Enterprise, and Edu users. Free, Plus, and Pro users will need to turn off the “Improve the model for everyone” setting to stop their data from being used to train the AI.
This tiered privacy model acknowledges different user needs. Enterprises handling sensitive data receive automatic protection, while individual users maintain control through settings. OpenAI states that projects inherit your existing privacy settings — and that no data is used to train models if those settings are turned off.
The Strategic Implications and Looking Forward
These upgrades position ChatGPT Projects as more than an organizational tool—they create persistent AI workspaces that learn and evolve with their users. The combination of voice integration, contextual memory, and granular controls suggests OpenAI views Projects as fundamental to ChatGPT’s future rather than an ancillary feature.
Projects will roll out to ChatGPT Enterprise/Edu early next year, indicating institutional adoption remains a priority. As organizations increasingly integrate AI into core workflows, features like Projects become essential infrastructure rather than nice-to-have additions. This aligns with broader 2025 AI trends showing AI moving from experimental tools to operational necessities.
The trajectory is clear: ChatGPT Projects will likely gain real-time collaboration features, deeper third-party integrations, and possibly project templates for common use cases. The current updates lay groundwork for these expansions while addressing immediate user needs.
For professionals already invested in ChatGPT Projects, these features enable entirely new approaches to AI-assisted work. The question isn’t whether to adopt these features, but how quickly organizations can restructure their processes to fully leverage them.
As AI tools proliferate, ChatGPT Projects demonstrates that the winning platforms won’t be those with the most features, but those that best understand how humans actually work. By focusing on continuity, context, and control, OpenAI has created something increasingly rare in the AI space: a tool that enhances human capability without requiring humans to fundamentally change how they think.
