AI News: Huge Updates From Anthropic, OpenAI and Google
This week brought major AI app updates from OpenAI, Anthropic, and Google, with new models including Claude Opus 4.7. OpenAI enhanced Codex with computer control and image generation, while Anthropic improved Claude Code with parallel sessions and Google released new TTS capabilities.
Summary
The video covers a busy week of AI developments across major companies. OpenAI significantly upgraded Codex, transforming it into more of a 'super app' that can now operate computers alongside users, generate images with GPT Image 1.5, remember preferences, and handle ongoing tasks through background computer use with multiple agents working in parallel. The host demonstrates creating both a website and a desktop Connect 4 game through Codex's new capabilities.
Anthropic updated Claude Code to support parallel sessions across multiple repositories, along with an integrated terminal, in-app file editor, faster diff viewer, and expanded preview capabilities. Google made their desktop app available to Windows users globally and released Gemini 3.1 Flash TTS, a new text-to-speech model with controllable voice emotions and timing. They also updated Nano Banana with personal intelligence integration.
The most discussed new model was Claude Opus 4.7, which shows significant improvements in coding benchmarks (64.3% on SWEBench Pro vs 53.4% for the previous version), though it falls between the previous public model and Anthropic's unreleased Mythos preview. Other new models included the open-source MiniMax M2.7 and Alibaba's Qwen 3.6 35B.
Additional updates included Perplexity's Personal Computer feature for local machine integration, upcoming Canva AI 2.0 features, Microsoft's MAI Image 2 Efficient model, Midjourney v8.1, DaVinci Resolve 21's AI features, and various other developments. The week also saw the unusual pivot of shoe company Allbirds to AI, causing a 600% stock jump, and an impressive Boston Dynamics robot demo completing household tasks from a whiteboard to-do list.
Key Insights
- The host argues that Codex is evolving into OpenAI's super app, consolidating chat, image generation, and code generation into one platform that can now operate computers with multiple agents working in parallel
- The speaker claims that Claude Opus 4.7 represents a significant coding improvement (64.3% vs 53.4% on SWEBench Pro) but most everyday users won't notice major differences unless they're coders
- The host demonstrates that modern AI coding tools can now build, test, and even play games autonomously, with Codex creating a Connect 4 app and then playing it to test the user experience
- The speaker believes the real game-changing AI use cases will come in biology, chemistry, and material sciences rather than productivity improvements, citing OpenAI's new GPT Rosalind for life science research
- The host suggests that newer AI models are getting better at understanding user intent with less prompt engineering required, meaning users need fewer attempts to get desired results from their conversations
Topics
Transcript
[0:06] It's been another huge week in the world of AI. As kind of with every week, we got big updates out of OpenAI, out of Google, out of Anthropic, out of Perplexity. And I'm here to break down what happened and what I think you should know about what happened in the world of AI this week. I'm not going to waste your time. Let's get right into it. Now, the news you're probably going to hear the most about this week from the most other YouTube channels is Claude Opus 4.7, the brand new model out of Anthropic. I do want to talk about this model, but it's not actually where I want to start today's video. We'll…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Matt Wolfe
AI News: GPT-5.6 and the new Super App are a Massive Leap!
OpenAI released GPT-5.6, a major leap forward in AI capabilities, alongside the new unified ChatGPT work app integrating code, browsing, and agent features. The week also saw significant model releases from xAI (Grock 4.5), Meta (Llama Spark 1.1), and research from Anthropic on AI reasoning patterns, establishing a new competitive landscape in AI development.
I Built A Monetizable Business With AI
The creator built a monetizable finance dashboard in one day using Hyper Agent from Airtable, which deploys a team of AI agents that continuously monitor news, track funding rounds and IPOs, analyze market sentiment, and generate daily briefings without manual intervention. The system demonstrates how specialized AI agents can work together autonomously to create a functional business product.
The ONLY AI Benchmark You Need!
A developer created "Buccy Bench," a humorous yet functional AI benchmark that tasks different language models with drawing Gary Busey as SVG code rather than images. The benchmark tracks model evolution over time while measuring performance metrics like cost, tokens, and execution speed.
GLM-5.2 - The Open Model That's As Good As Opus!
A comprehensive review of GLM-5.2, an open-weight Chinese AI model with a 1 million token context window, demonstrating its capabilities for coding, document analysis, and agentic workflows at significantly lower costs than frontier models like Claude Opus and GPT-4.5. The speaker tests various use cases including website building, Chrome extensions, game development, and data organization, concluding it's valuable for long, code-heavy, token-expensive tasks despite not universally outperforming closed-source alternatives.
Don't Fall For This AI Trap
The speaker emphasizes that power users distinguish themselves by knowing what NOT to automate with AI, rather than automating everything. They argue that AI works best for clear, straightforward tasks but struggles with nuanced, artistic work requiring consistency—using their failed YouTube thumbnail automation as an example.