Grok 3.5 is officially not launching this week. Elon Musk confirmed it’s “still too rough around the edges,” needing at least another week. With Gemini 2.5 Pro topping coding charts, it’s no wonder xAI doesn’t want to release something that doesn’t hit #1.
Meanwhile, Google is upgrading how we shop. Just three photos of a product and their Veo video model creates a full interactive 3D visualisation—giving even tiny sellers the tools of a Hollywood studio.
Higgsfield? They’re on fire—again. Effects Pack 3 just landed with cinematic effects like Morph Skin, Glow Shift, and even one so gory—Head Explosion—they had to censor the promo. And on a totally different front, an X user showed off real-time image recognition running locally on an M3 MacBook using SmolVLM. Fast, efficient, and no cloud required.
As Google I/O 2025 looms, Android 16 previews are here. Gemini AI will be everywhere: integrated into phone cameras, wearables, Android Auto—you name it. Gemini Live Assistant is going full omnipresent.
In the “adorable and terrifying” department, Baby Sam Altman popped up to explain how different generations use ChatGPT, and let’s just say: Gen Z’s brain is basically API-native now. Meanwhile, mem0 (or is it memO?) launched Open Memory MCP, giving AI agents memory across platforms. Expect smarter, more cohesive AI that remembers you across services.
Still no Grok 3.5? No worries. Tesla Optimus is making headlines with untethered dance moves. It even busted out some ballet. Yes—ballet. What once looked fake now looks like the future.
Over at OpenAI, GPT-4.1 is finally live inside ChatGPT—not just the API. It brings improved coding performance, now rolling out to Plus, Pro and Teams users. At the same time, Perplexity is turning into your next travel agent. Its new Travel Experience will search the web, build an itinerary, then—thanks to a PayPal/Venmo partnership—let you book everything directly from chat. Shopping and finance tabs are coming too. Google, watch your back.
CodeRabbit joined the race by integrating its real-time AI code review tool with VS Code, Cursor and Windsurf. The promise? Faster, smarter, more secure code with less dev effort. Back to Tesla Optimus, it wasn’t done flexing. Today it returned with even more dance styles, cementing its lead in the humanoid robot race.
Then things got wild. Google DeepMind unveiled AlphaEvolve—a Gemini-powered coding agent capable of discovering new algorithms, solving geometry problems, improving chip design, and optimizing training processes. In 20% of cases, it improved on known best solutions. That’s not just machine learning—that’s a taste of AI singularity.
Also this week: Gemini Advanced gained GitHub integration. You can now paste a repo URL and get function generation, debugging, code explanations and more, making Gemini an even more powerful AI developer assistant.
ElevenLabs turned up the creativity with their SB1 Infinite Soundboard. Just type a prompt, and boom—custom sound effects mapped to a digital drum pad. Mix, loop, and build your own FX machine in seconds.
And in China, robot warfare is no longer science fiction—it’s MCA: Mech Combat Arena, where bots will duke it out on TV later this month. The future isn’t coming. It’s already fighting.
Back to OpenAI—this week they launched Codex, an autonomous coding agent powered by their new model Codex One. It doesn’t just assist—it writes, debugs, and handles multiple software tasks on its own. And yes, this comes hot on the heels of their $3 billion Windsurf acquisition. The war for AI dev dominance is officially heating up.
Higgsfield wasn’t done either. Their new tool Higgsfield Ads turns a single photo into a professional-quality ad, with 40+ templates and 80+ motion styles. Perfect for startups wanting to look like Apple on a shoestring.
And to cap it all off? Tesla’s Full Self-Driving system just conquered the Arc de Triomphe roundabout in Paris—widely regarded as one of the world’s toughest driving environments. If FSD can handle that, what can’t it do?