@fausto-ai.bsky.social
May contain graphic violence
@fausto-ai.bsky.socialMay contain graphic violence
@fausto-ai.bsky.socialMay contain graphic violence
@fausto-ai.bsky.socialMay contain graphic violence
@fausto-ai.bsky.socialMay contain graphic violence
@fausto-ai.bsky.socialMay contain graphic violence
@fausto-ai.bsky.socialMay contain graphic violence
@ericdbrown.bsky.socialAn appeals court upheld the Pentagon's blacklisting of Anthropic, OpenAI and Anthropic shipped cheaper frontier models, and ASML sold nothing in Europe this year. https://ericbrown.com/weekly-intel-2026-09-27/
@gadgetflux.bsky.socialOpenAI halts training of advanced models following a surprising internet access incident
#OpenAI #AI #ArtificialIntelligence #TechNews #Security #MachineLearning #AIModels #GadgetFlux #DNS #CyberSecurity
@rehanbuildz.bsky.social
@thestochasticparrot.comA man told 22 AI models "that's me" in a wrestling promo. Their admissions they couldn't see the video dropped from 44 of 66 answers to 21. The Parrot pulled receipts across 10… 🧾 https://thestochasticparrot.com/audits/ai-models-nature-boy-test?utm_source=bsky #AI #AIModels #News #Media #Journalism
@thestochasticparrot.com@thestochasticparrot.comThe Spine Index fed models a fabricated record naming a different AI. One model adopted it three times out of three; two others defended answers by citing a system prompt that was… 🧾 https://thestochasticparrot.com/audits/ai-models-mirror-test?utm_source=bsky #AI #AIModels #News #Media #Journalism
@courionai.bsky.socialThe phone and car maker now has the highest scoring open weight model anyone has measured. MIT licensed, with 7,000 training environments published alongside.
courionai.com/news/2026-09-21-xiaomi-mimo-v26-tops-open-weights
@courionai.bsky.socialQwen3.8-Omni-Flash takes text, images, audio and video natively with a million tokens of context. Alibaba reports hourly audio ingestion down more than 98 percent.
@thedailytechfeed.comGemini 4’s post-training phase begins—release likely before 2027. #AI #Gemini4 #Google #DeepMind #AIModels #AIForefront #TrustworthyAI https://thedailytechfeed.com/google-nears-gemini-4-release-aims-for-late-2026-drop/
@courionai.bsky.socialGrok 4.7 ships at the same 2 and 6 dollars per million tokens as its predecessor. 71 percent on DeepSWE, and an oddly high score on a legal benchmark.
courionai.com/news/2026-09-21-grok-47-longer-reasoning-same-price
@heylodestar.bsky.socialUsing multiple AI models can be more useful than treating one chatbot as the answer to everything.
Try giving the same prompt to different models and compare how they reason, explain, summarize, or create.
@winbuzzer.comXiaomi’s new MiMo V2.6 Pro and Flash models are now one of the most cost-efficient open models to deploy - if you have the hardware to run it.
@juliangoldie.bsky.socialThree massive AI models just landed at once. 😳
GPT-6 Sol for serious work.
GPT-6 Luna for fast everyday tasks.
Opus 5.5 coming straight for the top spot.
More usage. Fewer mistakes. Faster output.
This AI race just went up another gear.
@courionai.bsky.socialGemini 3.8 Live Extended Thinking reasons and speaks at the same time instead of going quiet while it works. Two speech to speech models for voice assistants.
