JoeKnowTech

TL;DR: OpenAI quietly cancelled GPT-6.1 Astra after undisclosed safety failures, then released GPT-6.1 Terra at $1/$5 per million tokens. October 2026 was one of the busiest AI months ever, with major drops from Meta, Mistral, Anthropic, Google, and xAI within weeks of each other. So OpenAI had a big one cancel this month. GPT-6.1 Astra got pulled before launch. The official word is vague: safety evaluation findings that require additional alignment work. That sounds calm and corporate but probably means something genuinely concerning happened during red-teaming. And then almost immediately, OpenAI announced GPT-6.1 Terra. Same generation number, completely different positioning. At $1 per million input tokens and $5 per million output, it slots in below GPT-6 Opus in the pricing stack. This is OpenAI doing triage in public. What Actually Happened With Astra? OpenAI has not been specific, and that is intentional. Safety evaluation findings tells you something went wrong witho...

TL;DR: Apple unveiled its first-ever touchscreen MacBook Pro on October 28, with the 14-inch starting at $2,899 and the 16-inch at $3,399. It features a PROLED XDR display, Dynamic Island cutout, M5 Pro or M5 Max chips with up to 128GB RAM, and goes on sale November 6. Preorders opened October 30. I genuinely did not think Apple would ever put a touchscreen on a Mac. They spent years telling us why touch on a laptop was a bad idea. And then, just like that, they did it. The new MacBook Pro launched October 28, and it's a bigger deal than the usual spec bump. This isn't just a touch layer slapped onto the existing display. Apple redesigned the whole thing from the hinge up to make touch actually work on a laptop. What You're Actually Getting The display is what Apple is calling PROLED XDR. Not OLED, not mini-LED as we knew it, but a new hybrid panel that hits 2,000 nits for both HDR and SDR content with a 10,000,000:1 contrast ratio. The Touch Bar is gone (good rid...

TL;DR: Google just launched a satellite called Suncatcher that runs Gemma 4 AI models directly in orbit using custom TPU chips. The satellite processed its first real-time weather prediction in space this week, cutting latency from 40 seconds to under 3. This is either the beginning of edge AI in orbit or the most expensive demo Google has ever built. I've been following the Suncatcher project since Google first hinted at it in early 2026, and honestly I didn't think they'd pull it off this fast. But they did. Last week Google confirmed that Suncatcher, its experimental AI satellite, ran its first successful inference workload in orbit. Not on the ground. Not in a data center simulating orbit. Actually in space, 550 kilometers up, processing real atmospheric data with Gemma 4 running on radiation-hardened TPUs. What Is the Suncatcher Satellite Suncatcher is Google's first purpose-built AI inference satellite. It launched earlier this year on a SpaceX rideshare...

TL;DR: Intel shares dropped 29% on October 30, 2026 after guiding Q4 revenue to $15 to $16 billion, far below the $17 billion analysts expected. This dragged the entire tech sector down, with Nvidia off 3%, Tesla down 2%, and Alphabet continuing its freefall after losing nearly $500 billion in market cap the day before. The core fear: tech giants are spending $900 billion on AI this year and nobody knows when the returns will arrive. I have been watching tech earnings cycles for years, and yesterday was something I will not forget for a while. Intel did not just miss. It fell off a cliff. And it took the rest of the market with it. This was not just a bad quarter. This was a market saying: we are done waiting. What Intel Actually Said Intel guided Q4 revenue to $15 to $16 billion. Wall Street wanted $17 billion. That $1 to $2 billion gap does not sound massive, but in stock market terms it was a bomb. The company described its R&D and manufacturing spending as "signi...

TL;DR: Canada's federal government just took a C$1.6 billion equity stake in Cohere, an enterprise AI company most people outside Toronto have never heard of. The deal comes with strings: Cohere must keep its headquarters in Canada and commit to investing C$1.7 billion in Canadian computing infrastructure by 2030. This is the largest government AI investment in Canadian history, and it signals that Ottawa is done watching Silicon Valley set the terms for AI development. I had to look up Cohere twice when this news broke. Not because the company is obscure in AI circles, but because the scale of this bet is so far outside what governments usually do that I wanted to make sure I read the numbers right. C$1.6 billion. As an equity stake. In a private AI startup. That is not a grant. That is not a tax incentive. That is the Canadian government becoming an actual shareholder in an AI company, with all the upside and downside that comes with it. What Is Cohere? Cohere is an ...

TL;DR: Mistral AI just released the preview of Mistral Large 4, a one-trillion-parameter multimodal AI model trained in European data centers on nearly 4,000 Nvidia Grace Blackwell GPUs. Open weights are dropping by end of October 2026, making this one of the most capable open-source AI models ever released. Benchmarks show it beating GPT-6 Astra on legal and financial tasks and solving 93 percent of cybersecurity challenges. I have been following open-source AI for a while now, and this week had one of those moments where I actually stopped and re-read the headline twice. Mistral just released a preview of their latest model and the specs are genuinely wild. One trillion parameters. Multimodal. Open weights. Made in Europe. That combination does not exist anywhere else right now, and I think a lot of people are sleeping on what this means for developers, researchers, and businesses who cannot afford to pay per-token API bills forever. What Is Mistral Large 4? Mistral Larg...