skip to content
Site header image Minh Ton

This Week in AI #6 - Feb 24 to Mar 1

A roundup of all the latest AI news and releases from Feb 24 to Mar 1, 2025.

Last Updated:

👋
Hi there! This Week in AI is a weekly series of notes where I try to stay updated with the latest news in Artificial Intelligence!

My Favorites This Week

OpenAI GPT-4.5

Feb 28, 2025

GPT-4.5 is OpenAI’s latest and biggest model for chat. It features a broader knowledge base, improved user intent understanding, and enhanced emotional intelligence. It excels in writing, coding, and problem-solving, with reduced hallucinations.

OpenAI Introducing GPT-4.5

Claude 3.7 Sonnet & Claude Code

Feb 25, 2025

Anthropic’s Claude 3.7 Sonnet is a hybrid reasoning AI, producing near-instant responses or extended, step-by-step thinking.

AnthropicAI Claude 3.7 Sonnet and Claude Code

Claude Code, a command-line tool for agentic coding is also released.

Other Releases

Pika 2.2 & Pikaframes

Feb 28, 2025

Pika 2.2 introduces Pikaframes, a new feature enabling smoother, more controlled video transitions. It enhances image-to-video, text-to-video, and Pikascene generations for more natural, extended content flow.

Pika Labs AI pikalabsai.org Pika 2.2: Extend Video using Pikaframes available Now

HumeAI’s Octave

Feb 27, 2025

Octave is a next-gen speech-language model with emergent capabilities, including real-time voice and personality creation.

Twitter Hume on Twitter / X

Microsoft’s Phi 4 Multimodal

Feb 27, 2025

Phi 4 Multimodal processes text, images, and audio, generates text outputs, and features a 128K token context length for enhanced comprehension.

Microsoft Azure Blog Weizhu Chen Empowering innovation: The next generation of the Phi family | Microsoft Azure Blog

Alexa+

Feb 27, 2025

Alexa+ is a next-gen AI assistant that dynamically switches between LLMs like Nova and Claude, enabling smarter task handling, bookings, and purchases.

Panos Panay, SVP of Devices & Services Introducing Alexa+, the next generation of Alexa

ElevenLabs’s Scribe

Feb 26, 2025

Scribe is the most accurate Speech to Text model. It outperforms Gemini 2.0 and Whisper v3, setting a new benchmark for speech-to-text accuracy. It supports 99 languages, speaker diarization, character-level timestamps, and non-speech events like laughter.

ElevenLabs ElevenLabs — Meet Scribe the world's most accurate ASR model | ElevenLabs

Gemini Code Assist

Feb 26, 2025

Gemini Code Assist offers AI-powered coding tools and code reviews, enhancing development efficiency and accuracy.

Google Cloud Gemini Code Assist for business

Alibaba’s Wan 2.1

Feb 26, 2025

Wan 2.1 is a cutting-edge open-source video model that surpasses top models like Sora in benchmarks while generating videos much faster. It supports text-to-video, image-to-video, and video-to-audio.

Twitter Wan on Twitter / X

Qwen QwQ-Max-Preview

Feb 25, 2025

QwQ-Max-Preview is a reasoning model based on Qwen2.5-Max, offering enhanced math, coding, and agent capabilities. It is smarter and more creative but remains in the preview stage.

Qwen Qwen Team <think>...</think> QwQ-Max-Preview

Kaiber AI’s SuperStudio Pro

Feb 23, 2025

Kaiber Superstudio from Kaiber

Kaiber AI’s SuperStudio Pro is a powerful node-based platform for generating storyboards, videos, voiceovers, SFX (MMAudio), and more. It also lets users train characters, objects, environments, and styles in one seamless interface.



👏 Thanks for reading!