All AI news

AIVision is coming to both soon.

AITetris: adding the screen raises cleared lines from 70 to 81 > Wordle: solved 12/12 games in 3.8 guesses on average, reading the board from screenshots. No text version of the Wordle board needed. 3/4

AILiquid AI released d1 with vision support, accepting images, text, or both as inputs. In tests on six real applications, d1 matched or beat GPT-6.1 Sol on four while costing 19x to 200x less than both GPT-6.1 Sol and Claude Opus 5.5. It returns probabilities for yes/no, choice, or score questions in one forward pass, with text decisions in 200 to 300 ms.

AIIt is currently the best vision-enabled decision model on the market. > 85% to 97% accuracy across four VisA inspection tasks > circuit boards, candles, cashews, and chewing gum > understands the task from a short description, without task-specific training 2/4

AIevery release since launch with what's new, improved and fixed. there's also an rss feed if you want to access it programmatically

AIThe real work is getting new capabilities into the hands of people making things, then making sure they can actually use them. That’s how research becomes part of the craft.

AI…blockers and nitpicks - Codex is usually LGTM 👍 So either Opus 5.5 is the best coder right now or GPT 6.1 Sol is just super chill

AI…Qwen 3.8 27B running locally in both reasoning and non-reasoning modes

AIThe article examines whether multi-agent swarms could become a new scaling law, comparing them with inference scaling from o1. OpenAI researcher Noam Brown said its models are now sometimes trained with other agents, while the cited Anthropic data suggests gains beyond 10 agents are smaller and mainly speed-related. The article also raises the risks of groupthink and misaligned agents, and it notes that a Microsoft Research and UC Berkeley paper found teams sometimes solved tasks solo agents could not.

AILiquidAI has released d1-omni-600M on Hugging Face, a 587M-parameter model that answers named yes/no, choice and score questions over text, images or up to 30 seconds of speech in a single forward pass. It returns typed answers with zero output tokens by reading the model's distribution over options, and is built on LFM2.5-Encoder-350M with a 16,384-token context length. The model is not a chat model and does not generate text.

AIOtherwise, you get the same AI slop as everybody else that something might not always be the code until one day the model is 10x/100x smarter than us in every dimension, and you actually get worse results by interfering at which point why do anything at all. plug me into the matrix and let me live out my days in 1990s chicago

AI…singularity is confusing, it's possible to imagine things beyond it - like heists.

AIMAI-Transcribe-2-Streaming hears you • Gemma 4 on LiveKit Inference reasons and calls tools • MAI-Voice-2.1-Flash answers What do you think? STT: TTS:

AILiving Models paired Google's Gemma 4 with BOTANIC-1, a plant-DNA-trained AI, to identify the mutations behind crop traits. In one melon-yield test, the pipeline ranked the target mutation first among 2,494 candidates in under four minutes. The approach aims to speed development of climate-resilient crops that traditionally take years of field trials.

AIThe multiyear agreement covers cross-licensing in 5G, computing, AI and networking. Qualcomm will also buy some of Huawei's U.S. patents: ownership, not just permission to use them. Financial terms haven't been disclosed, and the transaction still needs regulatory approval.

AI…non-resident senior fellow.

AITwo new studies from Stanford researchers and collaborators, supported by Stanford HAI, ask whether those tests measure what they claim to. Read more:

AIBackground subagent results come back locally in TypeScript and Python. See everything new in the SDK:

AITheir results return to the parent as a follow-up turn on the same run, so stream() and wait() carry through until every subagent finishes.

AIrun.steer() adds your message to the next turn. If a subagent is mid-task, it moves to the background and keeps working.

AIZero margin for error. AWS & @f1 built an agentic AI solution that resolves critical issues up to 86% faster, so engineers focus on the future, not the logs.

AIRun frontier open models in your favorite coding harness. Track your spending and let our Auto router pick the model for you: low-cost models for quick fixes, more capable models for harder tasks.

AI…Artificial Analysis Intelligence Index score of 24 - above models with 10x the active parameters

AInow Krea Agent can create motion graphics, resize your videos, and handle sound, images, and more, all inside After Effects. try it now for free 👇

AIWhen more and more business knocked on the door at Cockle Finance, Dan and his dad, Steve, needed a plan to keep up. So they made their own tool and solved a beautiful challenge, taking on more business than they ever could before. Custom tooling, built simply, allowed them to scale. Made by Dan. Made on Replit.

AIJosh A. Goldstein of CSET has commented in several media outlets on AI-enabled influence and spam, including Tech Policy Press, MIT Technology Review, NPR, and the Financial Times. His commentary covers Meta's quarterly threat report on five fake-account networks linked to Moldova, Iran, Lebanon, and India, and OpenAI's first report on misuse of its generative AI. He also discussed a surge in AI-generated spam on Facebook and other platforms.