🚀 GPT-4.5 is here!

Also: Amazon announces Alexa+ with Generative AI

Welcome, AI enthusiasts

OpenAI has unveiled GPT-4.5, its most knowledgeable AI model yet. It comes with major improvements, offering a 10x boost in computational efficiency. That means more natural writing, sharper pattern recognition, and fewer AI hallucinations.

Let’s dive in!

In today’s insights:

  • GPT-4.5 is here!

  • Amazon announces Alexa+ with Generative AI

  • Alibaba open-sources Wan 2.1 AI video models

Read time: 4 minutes

LATEST DEVELOPMENTS

Evolving AI: OpenAI launches GPT-4.5, its largest model yet, improving efficiency and performance without being classified as a frontier AI.

Key Points:

  • OpenAI’s GPT-4.5 is now available as a research preview for ChatGPT Pro users, with a broader rollout planned.

  • The model is more efficient than GPT-4, offering improved writing, programming, and problem-solving abilities.

  • GPT-4.5 does not meet OpenAI’s criteria for a frontier model but aligns better with human interaction and reasoning.

Details:

OpenAI has announced GPT-4.5, describing it as its most knowledgeable AI model to date. While not considered a frontier AI, it improves upon GPT-4 with a 10x boost in computational efficiency. Users can expect a model that writes more naturally, recognizes patterns better, and hallucinates less. OpenAI’s new supervision techniques, combined with traditional fine-tuning and reinforcement learning, have helped shape its refined capabilities.

Why It Matters:

GPT-4.5 offers immediate benefits for writers, developers, and researchers looking for a more reliable AI collaborator. Its ability to generate text with fewer hallucinations makes it more trustworthy for professional use. While it may not outperform OpenAI’s reasoning-focused o1 and o3-mini models, its efficiency improvements signal progress toward making AI more practical for everyday tasks. With GPT-5 on the horizon, OpenAI continues its push toward more integrated, capable AI systems.

Evolving AI: Amazon unveils Alexa+, a generative AI-powered assistant with personalisation and agentic abilities.

Key Points:

  • Alexa+ enables smoother conversations and learns user preferences over time.

  • The assistant can handle tasks like booking rides, making reservations, and summarising documents.

  • Amazon employs multiple AI models, including Nova and Claude, for optimal performance.

Details:

Amazon introduced Alexa+ at its Devices and Services event, integrating generative AI to improve interactions. The assistant now offers more natural conversations, remembers user habits, and handles complex tasks independently. It can process documents and emails, pulling key information to create summaries or schedule events. Users can also control smart home devices, transfer music playback, and jump to specific scenes in Prime Video. Alexa+ will launch in the U.S. on select Echo Show devices, with Prime members receiving access at no extra cost.

Why It Matters:

Alexa+ shifts from being a simple voice assistant to an active digital helper. With the ability to manage bookings, analyze emails, and process documents, it streamlines daily tasks. This seems to be what Apple’s Siri always promised to become. The question remains: Is Amazon too late to the AI race, or is Apple's Siri frighteningly behind?

Evolving AI: Alibaba has released Wan 2.1, a suite of open-source AI video generation models claimed to outperform OpenAI’s Sora.

Key Points:

  • Wan 2.1 includes four models capable of generating realistic videos from text and image prompts.

  • The smallest variant, T2V-1.3B, runs on consumer-grade GPUs with minimal VRAM.

  • The models leverage a novel 3D causal variational autoencoder for improved efficiency and quality.

Details:

Alibaba’s Wan 2.1 AI video generation models are now available on Hugging Face under an open-source Apache 2.0 license. These models introduce Wan-VAE, a 3D causal variational autoencoder that enhances spatiotemporal compression and reduces memory load. The open-source versions focus on video generation but lack video editing and video-to-audio conversion capabilities. Alibaba claims Wan 2.1 surpasses OpenAI’s Sora in consistency, scene generation, and spatial accuracy.

Why It Matters:

While the Wan 2.1 suite is aimed at video generation, they can also perform other functions such as image generation, video-to-audio generation, and video editing. Truly a big step forward for creatives by Alibaba to making some of the best AI tools available to everyone.

 👀 Click on the image you think is real

QUICK HITS

🗣️ ElevenLabs is launching its own speech-to-text model.

🚀 PikaLabs released Pika 2.2 with 10s generations, 1080p resolution.

👀 Ideogram released it’s latest model with speed and photorealistic AI graphics.

🌐 Perplexity teases Comet, an AI browser for ‘agentic search’.

😮 Hume’s new text-to-speech model Octave that understands what you’re saying.

📈 Trending AI Tools

  • 🌍 DupDub - AI Dubbing tool that automates video translation (link)

  • 🔧 theGist - Unified AI workspace to supercharge your enterprise (link)

  • 📸 Headshot Pro - Get professional headshots in minutes (link)

  • 🚀 Startilla - AI-powered business development assistant (link)

  • 🔍 Glimpse - AI-powered tool designed to streamline online research (link)

Reply

or to participate.