12 Days of OpenAI: Updates Every Interior Designer Should Know
OpenAI’s 12 Days of OpenAI introduces a range of tools and updates that are set to transform industries, including interior design. For those seeking smarter ways to streamline workflows or enhance creativity, these advancements offer exciting possibilities.
From advanced reasoning models like 01 and 03 to tools for video generation and collaborative project management, these updates tackle creative challenges with faster, more intuitive solutions. In this blog, we’ll explore how these innovations can empower your design practice, helping you save time, elevate creativity, and rethink how you approach your projects. AI might just be the creative partner we’ve all been waiting for.
Day 1: Meet o1 and o1 Pro
OpenAI kicked off its "12 Days of OpenAI" with two major announcements: the release of the 01 Model and the introduction of ChatGPT Pro.
01 Model: Smarter, Faster, Multimodal AI
The 01 Model is OpenAI’s most advanced release to date, offering:
Multimodal Capabilities: Handles text and image inputs seamlessly.
Enhanced Speed: Processes responses up to 60% faster than GPT-4.
Improved Accuracy: Reduces major mistakes by 34%, making it ideal for technical problem-solving, coding, and math.
Critical Thinking: Designed to analyze and respond more effectively to complex queries.
ChatGPT Pro: A Premium Experience
OpenAI also introduced ChatGPT Pro, a $200/month subscription tier designed for advanced users:
Unlimited Access: Includes full use of models like 01, GPT-4, and the specialized 01 Pro Mode for high-level tasks.
Advanced Features: Offers tools like voice mode and ensures reliability for critical challenges.
Professional Performance: Pro Mode demonstrates exceptional efficiency, significantly reducing time spent on detailed, technical problems.
Day 2: Reinforcement Fine-Tuning (RFT)
Day 2 introduced Reinforcement Fine-Tuning (RFT), a new approach to making AI models smarter and better suited for specialized tasks.
What Is Reinforcement Fine-Tuning?
RFT customizes AI models to handle specific challenges with expert-level precision.
How It’s Different: Unlike traditional methods, RFT helps AI think critically about unique problems rather than just following patterns.
Key Applications: Ideal for areas like law, healthcare, and engineering where specialized knowledge is essential.
Real-Life Examples
Legal Assistance: OpenAI partnered with Thomson Reuters to develop Co-Counsel AI, a legal assistant trained for complex tasks using RFT.
Scientific Research: A smaller 01 model fine-tuned with RFT successfully identified genes linked to rare diseases, outperforming its standard counterpart.
How It Works
Data Input: Provide task examples to train the model.
Grading Feedback: A grader evaluates the model’s responses, teaching it the correct reasoning paths.
Quick Adaptation: With only a few examples, the model can generalize and handle similar tasks efficiently.
Day 3: Enter Sora, the Text-to-Video Generator
Day 3 unveiled Sora, OpenAI’s innovative video generation tool designed to push creative boundaries and enhance AI-human collaboration.
Core Features
Sora Turbo: An upgraded version of the February model, offering faster speeds, affordability, 1080p resolution, and video lengths up to 20 seconds.
Creative Tools:
Storyboard: Plan multi-step video sequences on a timeline.
Remix: Alter existing videos with new styles or elements, adjusting remix intensity.
Loop: Create seamless repeating scenes by connecting start and end points.
Blend: Merge two videos into a unified, original scene.
Folders and Presets: Organize projects and save preferred styles for consistent results.
Community Feed: Explore user-generated videos and learn how they were created, fostering collaboration and inspiration.
Access and Availability
Access: Live in most countries, excluding Europe, the UK, and a few others.
Generations:
ChatGPT Plus users: 50 generations per month.
Pro users: Unlimited slow queue access and 500 fast generations monthly.
OpenAI’s Vision
Sora marks a shift from text-based tools to dynamic video creation, advancing co-creative AI capabilities. The product emphasizes safety and moderation while prioritizing creative expression. It also plays a key role in OpenAI’s broader goals, including exploring Artificial General Intelligence (AGI).
Day 4: Collaborate Smarter with Canvas
Day 4 introduced Canvas, a collaborative writing and coding workspace now integrated into ChatGPT. Canvas provides a side-by-side editor to streamline writing, coding, and document collaboration, bridging conversational AI with a structured workspace.
Key Features
Available to All Users: Canvas is now accessible to both free and paid ChatGPT users via the tool selector or automatically for relevant tasks.
Python Code Execution: Run Python code directly within Canvas, with syntax highlighting, autocomplete, a WebAssembly-based Python emulator for fast execution, and real-time debugging.
Writing Tools: Includes inline comments, tone and readability adjustments, spelling and grammar checks, and optional emoji additions for personalization.
Integration with Custom GPTs: Canvas supports Custom GPTs, enabling enhanced collaboration for tasks like writing letters or creating content.
Use Cases
Creative Writing: Draft and refine stories, essays, or letters with real-time ChatGPT suggestions.
Coding: Debug and run Python code, generate visual outputs like charts, and track changes within the editor.
Specialized Workflows: Custom GPTs can use Canvas for unique tasks, such as personalized content creation.
Day 5: ChatGPT Comes to Apple Devices
Day 5 brings exciting news for Apple users—ChatGPT integration is now available on iOS and macOS, making AI more accessible and versatile than ever. Here’s how this development could empower interior designers:
Key Features
Device Integration: ChatGPT is now accessible through Siri, Apple’s writing tools, and the camera on iPhone 16 for visual intelligence.
No Account Needed: Use ChatGPT anonymously or log in for personalized features and syncing across devices.
New Capabilities
ChatGPT via Siri:
Siri collaborates with ChatGPT to handle complex tasks, such as planning events or generating playlists.
Users can confirm requests before information is shared.
Writing Tools:
Integrated into Apple’s writing apps for tasks like summarizing documents, creating content, and refining text.
Advanced options include adding emojis or generating visuals like album covers.
Visual Intelligence with Camera:
On iPhone 16, ChatGPT can analyze images through the camera, identifying objects, evaluating details, or hosting creative contests like ranking festive sweaters.
On macOS
ChatGPT Extension: Integrated with macOS 15.2+ for use in apps like Preview and writing tools.
In-App Continuity: Conversations started on Siri or other apps can seamlessly continue in the ChatGPT app.
Day 6: Video, Screen Sharing, and Santa Mode 🎅
Day 6 introduces exciting new features that enhance user interactions with ChatGPT, including video, screen sharing, and a festive Santa mode!
Key Features
Advanced Voice Mode with Video & Screen Sharing:
Share live video and screens directly with ChatGPT for richer, more dynamic interactions.
Ideal for learning, troubleshooting, and collaboration.
Santa Mode for the Holidays:
Engage in real-time conversations with Santa, featuring his signature jolly voice.
Available across all supported platforms, bringing holiday cheer to your ChatGPT experience.
Feature Highlights
Video in Advanced Voice Mode:
Share live video with ChatGPT for interactive, real-time guidance.
Example: A user demonstrated how ChatGPT guided them step-by-step to perfect their pour-over coffee technique.
Screen Sharing:
Share your device’s screen for personalized help with tasks like crafting responses or troubleshooting.
Santa Mode:
Have fun, interactive chats with Santa—ask about reindeer, traditions, or even beard tips.
Includes holiday jokes like Santa’s favorite: “What’s every elf’s favorite music? Rap music!”
Access and Availability
Video and Screen Sharing:
Rolling out on mobile apps this week for Teams, Plus, and Pro subscribers.
Coming to Europe soon, with Enterprise and EDU plans launching in early 2024.
Santa Mode:
Available globally today across mobile apps, desktop apps, and the web.
Day 7: Introducing Projects in ChatGPT
Day 7 unveiled Projects, a powerful new feature in ChatGPT designed to revolutionize task management and organization. With tools that cater to diverse workflows, Projects empowers users to streamline productivity like never before.
Projects Overview
Projects act as dedicated workspaces where users can:
Organize Conversations: Group related chats in one place for easy access.
Upload Files: Attach relevant files, images, or spreadsheets.
Set Custom Instructions: Tailor ChatGPT’s behavior and responses to specific tasks.
Key Features
Seamless Organization:
Create, name, and color-code projects.
Drag-and-drop or search to add existing conversations.
Integrated Workflows:
Upload files and instructions to streamline responses within project context.
Example: Upload a home maintenance log and query past or upcoming tasks.
Dynamic Search and Canvas Integration:
Search past conversations or the web for project content.
Use Canvas to draft documents, emails, or code snippets directly in a project.
Availability
Now Available: Plus, Pro, and Teams users.
Coming Soon: Free-tier users.
Early 2025: Enterprise and EDU plans.
Day 8: Enhanced Search Features in ChatGPT
Day 8 spotlights significant improvements to ChatGPT Search, enhancing accessibility, performance, and user experience for all platforms.
Key Updates
Search for Free Users:
Search functionality is now available to all logged-in free users across platforms, including iOS, Android, and web apps.
Enhanced Search Features:
Faster performance.
Improved integration for mobile devices.
New map-related features for location-based searches.
Search in Advanced Voice Mode:
Voice-enabled search allows users to access real-time web information conversationally.
Rolling out over the next few days to enhance voice interactions.
Refined Web Navigation:
Users can set ChatGPT Search as their default search engine for faster access to specific websites like Netflix or booking platforms.
Optimized Mobile Search:
Rich visuals and seamless map integrations on iOS and Android.
Examples include exploring local dining options or receiving contextual refinements for queries.
Day 9: Developer Updates and API Enhancements
Day 9 celebrated developers and startups leveraging the OpenAI API, introducing powerful updates to streamline the development process and enhance capabilities.
Key Announcements
01 Model Out of Preview:
Now in production with advanced features:
Function Calling: Enables interaction with backend APIs for dynamic task execution.
Structured Outputs: Delivers responses in precise formats like JSON, ideal for data-driven applications.
Developer Messages: Allows hierarchical instructions to guide the model’s behavior.
Reasoning Effort Parameter: Optimizes compute usage based on task complexity.
Vision Inputs: Supports image-based tasks, useful in fields like manufacturing and science.
Improved Real-Time API:
WebRTC Support: Simplifies real-time voice interactions with low-latency protocols.
Cost Reductions:
GPT-4 audio tokens are now 60% cheaper.
GPT-4 Mini audio tokens are 10x cheaper.
Python SDK Integration: Provides seamless access to the Real-Time API.
Preference Fine-Tuning:
A new method, Direct Preference Optimization, allows developers to align models with user preferences, optimizing for tone, style, and helpfulness.
Available for GPT-4, with plans to include GPT-4 Mini.
New SDKs:
Official SDKs for Go and Java now support all API endpoints, joining existing Python and Node.js SDKs.
Simplified login and API key generation process for easier integration.
Day 10: Expanding Accessibility with 1-800-CHAT-GPT and WhatsApp Integration
Day 10 focused on accessibility, introducing two new ways to interact with ChatGPT: voice calls and WhatsApp messaging. These updates aim to make AI more accessible and intuitive by leveraging widely used communication channels.
Key Announcements
ChatGPT via Phone Calls (US-only):
Dial 1-800-CHAT-GPT (1-800-242-8478) for conversational AI through any phone, including flip phones and rotary phones.
Features include:
Free Access: 15 minutes of free calling per month without an account.
Extended Conversations: Sign in with a ChatGPT account for longer interactions.
Examples:
Answering road trip questions about the “Flintstone House.”
Explaining reinforcement fine-tuning in simple terms.
Offering language tips like greeting someone in Spanish.
ChatGPT on WhatsApp (Global):
Available for text-based conversations worldwide.
Features include:
Quick Adjustments: Refine responses or adapt to user preferences.
Efficiency: Runs on GPT-4 Mini for fast, streamlined communication.
Examples:
Creating a pasta recipe and offering vegan or meat-based variations.
Future Plans: Integrate richer features like image interactions, search, and ChatGPT account functionalities.
Day 11: Enhanced Productivity with ChatGPT Desktop Apps
Day 11 brought exciting improvements to ChatGPT’s desktop app functionality, transforming it into a powerful tool for productivity, collaboration, and task automation.
Key Updates
Native Desktop Apps for Mac and Windows:
Lightweight, efficient, and designed for focused multitasking.
A dedicated shortcut (Option + Space) enables quick access to ChatGPT while working.
Enhanced App Interactions:
ChatGPT now integrates directly with desktop applications, eliminating the need for manual copy-pasting.
Example: Writing and running Git commands in Warp Terminal and visualizing results with a holiday-themed bar chart using Advanced Data Analysis (ADA).
Expanded Application Support:
Works seamlessly with apps like Apple Notes, Notion, and Quip for collaborative content creation.
Example: Crafting a historical walking tour in Notion while fact-checking via web search and adjusting tone to match the user’s style.
Voice Integration for Hands-Free Use:
ChatGPT’s Advanced Voice Mode works directly with desktop apps for conversational task management.
Example: Refining a holiday setlist using voice commands, including fun interactions with a Santa-themed assistant.
Accessibility Enhancements:
On MacOS, ChatGPT uses accessibility APIs to “see” the content of your active applications, providing more accurate, context-aware assistance.
Why This Update Stands Out
These features position ChatGPT as more than just a conversational tool—it’s evolving into a fully integrated assistant capable of performing context-aware tasks across applications. From coding assistance in IDEs like Xcode and VS Code to managing data and creating content, ChatGPT enhances productivity, collaboration, and creativity.
Availability
MacOS: All features are live today.
Windows: Updates coming soon.
Day 12: OpenAI's Next Frontier Models - 03 and 03 Mini
Day 12 wrapped up OpenAI’s 12 Days of Updates with the announcement of two groundbreaking models, 03 and 03 Mini, setting a new standard for AI performance, safety, and accessibility.
Key Announcements
New Models:
03: OpenAI’s most advanced model, excelling in programming, mathematics, reasoning, and general intelligence.
03 Mini: A cost-efficient version of 03, offering comparable performance with lower latency and cost, ideal for scalable use cases.
Public Safety Testing:
Both models are open for external safety testing starting today, inviting researchers to assess vulnerabilities and enhance security.
Applications for safety testing close on January 10, 2025.
Deliberative Alignment:
A new training method leveraging advanced models to clearly define the boundary between safe and unsafe prompts.
Reduces refusals of valid queries while improving the rejection of unsafe ones.
Performance Highlights
Coding:
03 achieved a CodeForces ELO of 2727, a leap over previous models, rivaling elite human programmers.
Mathematics:
Reached 96.7% accuracy on the Amy benchmark, nearing perfection.
GPQA (PhD-Level Science Questions):
Scored 87.7%, surpassing human experts (average 70%).
Arc AGI Benchmark:
Set a state-of-the-art score of 87.5%, exceeding human-level performance in general intelligence tests.
Cost Efficiency with 03 Mini:
Offers adjustable reasoning effort (low, medium, high) for balancing cost and speed based on task complexity.
Matches the performance of 01 with significantly lower latency and cost.
Demonstrations
Self-Evaluating Models:
03 Mini executed Python scripts to evaluate its own performance on datasets like GPQA.
Adaptive Reasoning Effort:
Users dynamically adjusted reasoning time to prioritize speed or response quality.
Why This Matters
Frontier Models: 03 and 03 Mini redefine what AI can achieve in reasoning and decision-making, tackling challenges previously considered out of reach.
Safety First: The public safety testing program represents a major step in AI transparency and collaboration.
Accessibility: 03 Mini makes powerful AI capabilities more affordable, enabling broader adoption by businesses and developers.
What’s Next
03 Mini Availability: Expected by end of January 2025.
03 Launch: To follow after safety testing is complete.
Benchmark Development: Collaboration with the ARC Foundation to create next-generation benchmarks for future AI models.
The 12 Days of OpenAI concluded with tools that are set to make a significant impact on how we, as interior design professionals, approach our work. With models like 03 Mini bringing advanced reasoning at an accessible cost and features such as enhanced search and collaborative workspaces redefining project management, these updates are designed to streamline workflows and enhance creativity.
As we navigate an industry increasingly influenced by technology, OpenAI’s advancements provide the tools we need to work smarter, collaborate more effectively, and focus on what we do best—designing inspiring spaces. The possibilities for integrating these tools into our practices are both exciting and limitless.
This article was written by Jenna Gaidusek, founder of AI for Interior Designers™, and enhanced using ChatGPT.