Meta's Muse Spark 1.1 includes a new paid tier for developers. Developers are able to use Meta's model for free up to a point, then they'll be required to pay for access. Meta's API pricing is roughly 25% of the cost of other top competing models. Mark Zuckerberg has described Muse Spark as having state-of-the-art - or very close to it - agentic reasoning and tool use.

OpenAI has released its new frontier model family for ChatGPT Work, Codex, and the OpenAI API. The GPT-5.6 family has three tiers: Sol, the flagship model; Terra, a lower-cost everyday work option; and Luna, the fastest and most affordable model. The series features higher intelligence per token, lower estimated cost for complex work, and stronger agentic performance. There is a new 'ultra' setting that coordinates multiple agents across parallel workstreams for demanding tasks.

Humanoid robots controlled by skilled human surgeons recently removed the gallbladders from living pigs in an unprecedented medical experiment. The approach takes a fraction of the space needed in an operating room and is easy to deploy. The experiment used a Unitree G1 humanoid robot, which has a starting price of $13,500. Unitree's humanoid robots teleoperated by surgeons are still very much in the experimental phase, but the approach could eventually be used in smaller hospitals and clinics that lack the resources to install expensive surgical robots.

Apple has reportedly held meetings with PrismML about how it could use the startup's technology to run much larger AI models directly on iPhones. PrismML has managed to shrink down Alibaba's Qwen 3.6 model, which has 27 billion parameters, to run entirely on an iPhone Pro. Larger models running directly on iPhones would allow for more Apple Intelligence features to run on-device, which could reduce Apple's costs and further enhance user privacy.

GitHub had over 14,000 repositories across its primary internal GitHub organization. Less than half had clear ownership. This post discusses how GitHub gave every active repository a validated owner in under 45 days, archived the rest, and made ownership the foundation for everything that followed.

Agents are writing code faster than any human can review. Humans being involved in every code review is a bottleneck. Put humans outside of the code review loop by building a pipeline that delegates tasks to agents. This post looks at power workflow changes that PostHog uses to make reviewing AI-generated code fast without losing quality.

Netflix's subscriber engagement is showing signs of decline. While the company remains the industry leader among subscription-streaming services, shares are down more than 40% over the past 12 months. Executives have started discussing adding live channels that would continuously stream certain programs to bolster engagement. They are also looking at other sports events they can add to the streaming service.

A class at Brown University was giving take-home exams, and the professor suspected students were using AI to complete them. After the professor announced that the final exam would be in person, 18 students dropped the course, and nine others didn't attend the final exam. Of those 27 students, 22 had scored perfectly on the midterm exam. The average test score plunged from 96 to 48 for those who took the test.

Mark Zuckerberg is not convinced that AI technology will ultimately become a commodity as some companies are already gatekeeping aspects of their technology.

Plan A is a wish list, a positive vision, a road map for navigating a future where this AI thing turns out to be real.

John Ternus should return Apple's privacy policy to its 2014 clarity.

The competitive dynamics in the next phase of AI will likely be defined by the unit economics of data acquisition.

The current market dynamics point to a future in which frontier models move towards becoming commodity infrastructure.

Mitchell Hashimoto, the developer behind Vagrant, Packer, Consul, Ghostty, and Vouch, talks about terminals, Zig, and open source.