A hobbyist project called bootgpt packs a tiny character-level language model into a 512-byte MBR plus 23 additional disk sectors, running entirely in 16-bit real mode without an OS or C runtime. The 11.5KB int8-quantized model, trained on synthetic dmesg logs, generates fake Linux boot messages with invented hardware IDs and timestamps as the machine starts up. The creator shared build scripts, training code and a boot image that can be written to a USB stick to run on real BIOS hardware.
github.com
· 2026-10-31
Canva has announced it built its own vision-language AI model aimed at producing design work, rather than relying solely on third-party generative AI systems. The company is positioning the model as a way to address the recognizable, low-quality aesthetic commonly associated with AI-generated graphics, often called 'AI slop.'
fastcompany.com
· 2026-10-30
On DoorDash's Thursday earnings call, CEO Tony Xu told analysts that AI agents have generated only a few hundred orders since launching, calling it a 'very low use case' against the roughly 1 billion total orders the company processed last quarter. Xu said customers still prefer to order food directly through DoorDash rather than via outside AI tools.
arstechnica.com
· 2026-10-30
Tom's Hardware has released an explainer breaking down core terminology behind large language models, covering concepts such as models, tokens, weights, KV cache, quantization and inference. The guide is aimed at readers following the growth of local AI hardware, such as Nvidia's RTX Spark, and is intended to support the publication's ongoing AI coverage.
tomshardware.com
· 2026-10-30
Strava has expanded its AI assistant integrations to include Google Gemini, following a similar rollout with Anthropic's Claude months earlier. Subscribers who connect their Strava account to Gemini can ask natural-language questions about their training history, such as pace trends, race performance, or how strength training affects running, with the integration described as optional and read-only.
9to5mac.com
· 2026-10-29
A hobbyist built two small neural networks—a looped transformer 'world model' that generates game frames and rewards, and a lightweight MLP 'actor' that chooses actions—running together entirely on a Raspberry Pi Pico 2 W microcontroller. The world model, ranging from 26,000 to 131,000 parameters, predicts the next frame and reward from the current frame and action, while the actor (9,000-16,000 parameters) picks moves based only on generated frames, never touching the real game logic. The project also supports a streaming reinforcement learning agent that can learn to play while running inside the simulated 'dream' itself.
github.com
· 2026-10-29
Apple's iOS 27.2 update, released days after iPhone Duo debuted in iOS 27.1, adds a new Shortcuts automation that triggers actions when the iPhone Duo device is opened or closed. The update also expands workout automation to the Fitness app, lets app-closed automations report which app closed, adds web-based multilingual support for the 'Use Model' action, and improves the Shortcut Editor with new macOS layout options and direct name/icon editing on iOS and iPadOS.
9to5mac.com
· 2026-10-29
Google has launched Gemini 4 Carbon, its second new frontier AI model in a month following Gemini 4 Argon. The model improves on coding, agentic tasks, scientific reasoning and multimodal understanding, with Google saying its biggest gains are in cybersecurity, including more than doubling secure-code generation rates versus Argon. Carbon will first go to select cyber defenders before wider release, priced at $2 per million input tokens and $10 per million output tokens.
androidauthority.com
· 2026-10-28
Google has announced Gemini 4 Carbon, a frontier AI model released just two weeks after Gemini 4 Argon, which it says improves on coding, agentic tasks, scientific analysis and cybersecurity. The company reports gains such as doubled vulnerability-free code generation versus Argon and a 10x improvement in patch quality claimed by its Chrome team, plus an 89-point jump on the WebDev Arena leaderboard. Despite announcing the model, Google is not yet making it broadly available, instead pursuing a staged rollout.
arstechnica.com
· 2026-10-28
Google has opened a beta Connected Apps Platform that lets any developer or business submit their app to work within Google Search and Gemini. Developers integrate via a cloud-hosted MCP server or Android MCP using AppFunctions APIs, then register and test apps through the Google Connected Console, choosing whether the app appears in Search, Gemini, or both. The beta is currently limited to developers in the U.S., and includes tools for tracking traffic, latency, error rates, and maintaining verified branding.
techcrunch.com
· 2026-10-28
Artificial Analysis has released version 5 of its Intelligence Index, the widely cited independent metric ranking frontier AI models, following criticism of flaws in prior benchmark aggregation methods. The piece discusses how AA's index and competitors like Epoch AI's Capabilities Index attempt to distill model performance into a single comparable score, despite inherent trade-offs in how such scores are aggregated.
interconnects.ai
· 2026-10-28
Google announced Genie 4, a real-time world model that builds on Gemini Omni and can maintain a consistent simulated environment for up to two minutes, double its Genie 3 predecessor. Users can alter scenes in real time with natural language prompts, including weather, camera movement, and new objects or characters, while the model generates ambient sound and adaptive human dialogue. Genie 4 works alongside Gemini 3.8 Flash, which breaks down prompts into steps, and is currently limited to select trusted testers in a research preview.
androidauthority.com
· 2026-10-28