🧠 OpenAI unveils GPT-4.5 'Orion'

Sign up | Sponsor

Hola, AI fam 🤖

In today’s WakeTheAI edition:

• OpenAI unveils GPT-4.5 ‘Orion’

• Alibaba’s AI video model Wan 2.1

• Prompt: Prioritize Tasks With MoSCoW Method

• Google upgrades Sheets with Gemini AI

• Meta plans to launch a standalone AI app in Q2

• OpenAI’s Sora is now available in the EU, UK, and more

• Meta unveils Aria Gen 2 glasses with advancing machine perception

OPENAI
🧠 OpenAI unveils GPT-4.5 with higher EQ

Lazy Bits: OpenAI has unveiled GPT-4.5, its most advanced AI model to date, designed to enhance user interactions by providing more accurate and emotionally intelligent responses.

In-Depth Details:

• Enhanced Emotional Intelligence: GPT-4.5 is engineered to interpret subtle social cues and nuances, making interactions feel more natural and human-like.

  

• Reduced Hallucinations: The model significantly lowers the incidence of generating incorrect or nonsensical information, with hallucination rates dropping from 61.8% to 37.1%.

  

• Resource-Intensive Deployment: GPT-4.5 is a large and compute-intensive model, making it more expensive to operate compared to its predecessors. ​

• API Pricing Increase: GPT-4.5 is significantly more expensive than GPT-4o, costing $75 per million input tokens and $150 per million output tokens, compared to GPT-4o’s $2.50 and $10 per million tokens.

• Selective Availability: Initially, GPT-4.5 is accessible to ChatGPT Pro subscribers, with plans to extend access to Plus and Team users in the coming weeks.

Lazy Conclusion: OpenAI's release of GPT-4.5 marks a significant advancement in AI capabilities, focusing on enhancing the quality and naturalness of human-computer interactions. However, the substantial increase in API pricing compared to GPT-4o may impact its adoption among developers and businesses.

TOGETHER WITH GAMMA

The future of presentations, powered by AI

Gamma’s AI creates beautiful presentations, websites, and more. No design or coding skills required. Try it free today.

ALIBABA
🎥 Alibaba’s AI video model Wan 2.1

Lazy Bits: Alibaba has launched Wan 2.1, an open-source AI video generation model that claims to outperform OpenAI's Sora in scene quality, object accuracy, and spatial positioning.

In-Depth Details:

• Model Variants: Wan 2.1 includes four models—T2V-1.3B and T2V-14B for text-to-video and I2V-14B-720P and I2V-14B-480P for image-to-video generation.

• Performance and Accessibility: The T2V-1.3B model runs on consumer GPUs with 8.19GB VRAM, generating a 5-second 480p video in under four minutes on an Nvidia RTX 4090.

• Advanced Architecture: Uses a diffusion transformer with 3D causal VAE, improving memory efficiency and generating stable 1080p videos while retaining frame consistency.

• Benchmark Results: Alibaba claims Wan 2.1 outperforms Sora with 32% better scene consistency, 28% higher object accuracy, and 20% improved spatial positioning in internal tests.

  

• Open-Source Availability: Released under Apache 2.0, it is free for research but comes with restrictions for commercial applications.

Lazy Conclusion: China is rapidly advancing in AI, with Alibaba’s Wan 2.1 offering a strong challenge to OpenAI’s Sora and Google’s Veo 2. As competition heats up, AI video generation is becoming a battleground where open-source and proprietary models will shape the future of digital content creation.

Prioritize Tasks With MoSCoW Method

You can ask ChatGPT to act as your project management specialist, helping you prioritize tasks using the MoSCoW method for effective workflow and resource allocation.

ChatGPT will categorize tasks into Must-Haves, Should-Haves, Could-Haves, and Won’t-Haves, providing clear justifications for each.

The output will be structured in a markdown table format with actionable recommendations, ensuring a well-balanced and efficient project execution.

Prompt:

You are a project management specialist with expertise in prioritizing tasks using the MoSCoW method. Your task is to develop a structured framework to help categorize project tasks effectively, ensuring a balanced workflow and optimized resource allocation.

Key Focus Areas:

1. Defining Task Importance – Evaluate tasks based on urgency, impact, and necessity to project success.
2. Applying the MoSCoW Framework – Categorize tasks into Must-Haves, Should-Haves, Could-Haves, and Won’t-Haves based on priority and feasibility.
3. Clarifying Task Prioritization – Provide brief justifications for the categorization of each task.
4. Optimizing Project Execution – Ensure a well-balanced workload that aligns with available resources and deadlines.
5. Enhancing Efficiency – Recommend task management best practices for seamless execution.

Key Information About Me:

Project Name: [Insert project name]
Project Deadline: [Specify deadline]
Available Resources: [List tools, budget, personnel, or software available]
Project Goals: [Define key objectives and success criteria]
Task List: [List all tasks that need prioritization]

Output Requirements:

• Present the prioritized task list in a markdown table format with four distinct columns:

Must-Haves – Critical tasks necessary for project success.
Should-Haves – Important tasks that enhance the project but aren’t mandatory.
Could-Haves – Non-essential improvements that add value but won’t impact delivery if excluded.
Won’t-Haves – Low-priority tasks that can be eliminated or deferred.

• Ensure clear and structured reasoning for each categorization.
• Use bullet points to highlight execution strategies for Must-Have & Should-Have tasks.
• Ensure recommendations are realistic, practical, and tailored to real-world project constraints.

Result:

• Meta unveils Aria Gen 2 glasses, advancing machine perception, AI, and robotics with enhanced sensors, on-device AI, and new accessibility features.

• OpenAI's Sora is now available in the EU, UK, and more. ChatGPT Plus and Pro users in these regions can generate videos using the model.

• Google upgrades Sheets with Gemini AI, enabling faster data analysis, trend detection, and visualizations like heatmaps, now available for Workspace business users.

• Meta plans to launch a standalone AI app in Q2 to compete with ChatGPT, aiming to lead in AI by year-end. A paid subscription model is also being tested.

1. 🧑‍💻 Interpret: All-in-one AI notetaker and translator

2. ✍️ TheGoodAI: Write high-quality essays instantly with AI

3. 📈 Luppa: All-in-one AI marketing platform