SpaceXAI has released Grok 4.6, a model focused on long-running agents, coding, and knowledge work. The model surpasses Kimi K3 and ties with GPT-5.6 Sol Max on the Artificial Analysis Intelligence Index. It posts sizable gains over its predecessor across coding, terminal, knowledge-work, and agent benchmarks. The API starts at $2 per million input tokens and $6 per million output tokens, less than half of what GPT-5.6 Sol costs in standard mode.

The Pixel 11 series looks a lot like last year's lineup, but with some minor tweaks, a few new hardware extras, and even more Gemini-powered AI. The phones are joined by an updated Pixel Watch 5 and the Pixel Tag, Google's first in-house tracker device. The most notable hardware innovation for the new Pixel lineup is the return of a notification LED for the Pro phones. The ring of RGB LEDs is housed inside the flash assembly on the back of the phone and lights up only when the user is using Gemini or when there is an incoming call from specific contacts for both standard phone calls and voice calls in WhatsApp.

Northrop Grumman launched a Mission Robotic Vehicle (MRV) and three Mission Extension Pods (MEPs) into orbit on a SpaceX Falcon 9 rocket in July. The MRV is a powerful satellite equipped with two advanced robotic arms. MEPs are simpler satellites that are essentially modular propulsion systems. In 2027, the MRV will use its robotic arms to attach one of the MEP pods to a telecommunications satellite, which should keep it in orbit for years to come.

Tens of thousands of workers in India are being recruited to record their work in first-person. The footage is being fed into AI systems being built to teach robots how to do things. Robotics companies currently lack the data needed to train machines that interact with the physical world. Real footage of humans doing mundane tasks is in short supply because no one cared about it, until now.

What looks like a rendering bug is actually a clever JPEG decoding optimization in Chrome. Chrome delegates image decoding and rendering to Skia, which does not always decompress the full image and scale it afterward. It computes the closest fraction with a denominator of 8 and decodes the image at that scale, then scales the image further using a more traditional downsampling algorithm until it reaches the desired size. This is why small images can look thicker on some machines.

AI makes projects with weak engineering culture fail much faster. Bad engineers were always a liability, but there used to be a limit to how fast changes could be implemented. Before, there was a good chance someone would catch their bad decisions before they went too far. Companies can't afford to hire bad engineers anymore. The money will increasingly get funneled towards an increasingly small number of people who can actually be trusted.

The share of businesses using model serving platforms rose again last month. 6.1% of businesses using AI are now using these platforms. Adoption of OpenAI and, to a lesser extent, Anthropic, has slowed in recent months. Their growth has to come from existing businesses spending on AI, and those businesses are increasingly spending on open source.

Palantir is a vertically integrated outcome-based software solution provider. It offers tech consulting and software services with a unique outcome-based model that lets it bypass the traditional trade-offs of both traditional consulting and software companies. It is vertically integrated because it has an extreme elitist attitude towards using its own software and services, unlike the rest of the market, which is much more polyamorous. There's little hope of any company being able to catch up to Palantir as investors allow the company to do R&D over an extremely long and unspecified time horizon.

DeepSeek has set up an official social media account and posted job listings for a new team focused on developing AI agents.

Be intentional about when to use AI and when to leave it out, and try to be honest about what you're actually gaining and losing with your choices.

Resolving the issue was a massive, cross-functional effort that involved dozens of people, including Tailscale's engineering and support teams and the core maintainers of SQLite.

Input-based pricing charges users for the consumption of some resource, while output-based pricing charges based on a positive outcome.

GitHub, Vercel, and Replit are trying to rebuild themselves around solving hard problems in the software development process other than coding.

A look at the horizon problem, job losses vs new company creation, specialized AI, system skills, and more.