AI News

MAPLE: Manifold-Activated Pruning of Latent Experts for Sub-2-BPW MoE LLMs - AmSach/maple

LiveReview is the org-wide risk-aware AI code review harness for business-critical systems — combining risk scoring, review-priority ranking, and a 10-category risk taxonomy, enforced on every commit, self-hosted or air-gapped.

Rex is ready for public testing and we're beginning to invite our mailing list.

We post-trained Alibaba's Qwen3.8-27B to reinforce Western values on a Chinese-trained model, and measured what changed with an 803-prompt eval across seven value categories.

Perplexity is launching Teammates, always-on AI agents that can complete multiple tasks at the same time.

Anthropic is poised to become the first pure-play AI company to hit the public market at a time when the public is increasingly leary of the industry.

Cinnamon is an AI agent your whole household can share. It plans, books, watches and reminds, for you and up to five people you trust, and always asks before it acts.

Prove a prompt change made your AI agent better, with paired replay. - gael55x/promptcheckup

Text-to-Speech model with 36B parameters. Download, explore files and run it on Hugging Face.

Image-Text-to-Text model with 28B parameters. Download, explore files and run it on Hugging Face.

Nvidia is requiring Southeast Asian cloud providers seeking Nvidia Cloud Partner status to enroll in its Fleet Intelligence tracking program, giving the chipmaker greater visibility into where its AI GPUs are deployed. The move comes as Washington increases pressure over suspected chip smuggling into China, with investigations involving Megaspeed International and Super Micro highlighting the challenges of enforcing U.S. export restrictions

OpenAI has reportedly raised an additional $11 billion as part of a funding round that hasn't closed yet but is expected to in the coming weeks.

Sentient is an open source monitor for coding agents. It captures every action your agent takes, from file changes to shell commands and network connections, so you can reconstruct exactly what happened.

Run recurring web searches on a schedule and get only new results, with cited answers, sent to your webhook. Track company news, RFPs, and competitors.

How coding models respond when familiar engineering tasks ask for something different.

HyperFlow H3 Flash: Video Rebirth's 13.7B, 4- to 8-step compressed MiniMax-H3 for joint video and audio generation (INT8 and GGUF) - Video-Rebirth/hyperflow-h3-flash

Run untrusted code in an isolated, on-demand environment.

Shield AI is raising $2.5 billion at a $28.8 billion valuation to acquire radar maker Echodyne in a $5.3 billion deal expected to close in early 2027.

AJX (Agent Journey Experience) is an open-source skill and CLI for agent usability testing. Test your website, CLI, SDK, or API and find what gets in an agent's way.

Makes aws, az, gcloud, ansible, eslint, rubocop, composer and hundreds of other Python, Ruby, PHP and Node CLIs start several times faster: each stays loaded in a daemon that forks it on every call. A checks script: 24.0 s → 5.3 s, same output. No per-tool config, one binary, Linux and macOS. - anthonyGosme/speedrunner

Serve your models fast on your own infra in 3 steps. - meshapplied/goethe

Text Generation model with 36B parameters. Download, explore files and run it on Hugging Face.

S&P 500 companies lost a record $2.9 trillion (€2.6 trillion) in market value on Thursday. European shares opened lower on Friday as investors weighed whether Democrats’ US midterm wins could lead to tougher rules on AI data centres.

I’m not the sort of person who typically pays for a house cleaner. I’m too frugal, and I kind of enjoy putting on a podcast and going to town on my tub with a Magic Eraser. But when I saw an X post about a new humanoid-robot cleaning service in San Francisco, I immediately signed…