Startup Hark, founded less than a year ago by Brett Adcock, released Hark Pro to the public today with a free tier and a paid subscription for heavier use. The app presents a full-screen interface with a central chat window, a feed of suggested actions, and visual 'panels' showing dashboards for things users care about, and it can connect to email, calendars, hard drives and payment accounts to carry out digital tasks on a user's behalf.
techcrunch.com
· 2026-10-06
Cedana released a web tool that converts reported token-processing volumes into estimated GPU requirements. Users enter a token figure and select a model, and the calculator returns a GPU estimate with a confidence range, accounting for differences in hardware, model type, and workload such as chat or agentic coding. Default assumptions are sourced and can be replaced with custom measurements.
cedana.com
· 2026-10-06
Google will limit free Gemini app and website users to only the Flash-Lite model, its lowest-tier offering, cutting off access to Flash and Pro starting October 9. AI Plus subscribers ($5/month) will lose the Pro model but keep Flash and Flash-Lite, while AI Pro subscribers ($20/month) gain access to the Deep Think reasoning mode previously limited to AI Ultra tiers costing $100-$200 monthly.
zdnet.com
· 2026-10-06
Hackers targeted major South Korean banks using an open-source AI security agent to assist in breaching the financial institutions' systems. The incident marks a notable instance of AI tools being incorporated into a real-world attack on banking infrastructure.
fastcompany.com
· 2026-10-06
OpenAI reportedly plans to release on GitHub hundreds of mathematical solutions generated by an internal model it began training in August, including a claimed resolution of the Navier-Stokes Millennium Prize problem. Mathematicians who attended an August meeting with OpenAI say they asked the company to publish such findings as peer-reviewable papers rather than a bulk dump, a request they say was ignored. OpenAI spokesperson Lindsay McCallum says the company is consulting an advisory group at the Institute for Advanced Study and has not set a release date.
wired.com
· 2026-10-06
Wells Fargo removed 500 megawatts of expected capacity resales from its 2027 estimates for Meta, roughly $5 billion in revenue, saying the company will likely need that computing power for its Muse AI agent instead. The firm raised its price target on Meta to $1,000 from $796, implying about 35% upside from Monday's close.
cnbc.com
· 2026-10-06
Google unveiled EmbeddingGemma 2, a 740-million-parameter multimodal AI model built on the Gemma 4 architecture and released under an Apache 2.0 license, designed to run locally on devices like Pixel phones using minimal RAM. The company also launched Google AI Edge Foresight, a Mac app that uses the model to transcribe meetings, expand shorthand notes into polished text, and answer questions by searching local files, calendars, and Google Drive content—all processed on-device.
9to5google.com
· 2026-10-06
New research from IANS, discussed by senior director Nick Kakolowski, finds that most CISOs are using AI to automate routine security tasks rather than reduce staff. Instead, security leaders are reskilling junior employees and freeing senior professionals to focus on more complex problems, while also leveraging AI momentum to justify budget requests.
darkreading.com
· 2026-10-06
Google introduced Nano Banana 2.1, a new image generation model succeeding its prior versions. The company says the model improves on visual design, mask-based editing, subject consistency, and produces more natural-looking images, and has made it available for testing.
twitter.com
· 2026-10-06
Apple has pushed macOS 27.2 public beta 2 to users enrolled in its public beta program, a day after seeding developer beta 3. Build 26B5101f matches the developer release, and the main addition is support for five new Siri AI languages—French, Japanese, Korean, Portuguese, and Spanish—alongside English.
9to5mac.com
· 2026-10-06
French AI startup Mistral released a new open-weight model, which it says performs competitively against leading open-source models developed in China. The company positioned the release as part of its effort to offer a European alternative in the global open-source AI race.
news.slashdot.org
· 2026-10-06
A project called openTPU has released an open-source AI inference accelerator whose hardware design, instruction set, compiler and simulator were developed by AI agents rather than human engineers. The design was tested on an Inspur YPCB-00338 FPGA card, successfully running ten language models including LFM2.5, Qwen3, Gemma 4 and SmolLM3, with output matching the simulator bit-for-bit.
github.com
· 2026-10-06