MarkTechPost (4 hours ago)
Anthropic released FlashProve, an Apache 2.0 zkSNARK prototype that proves a reply came from a committed model without revealing weights. It proves Llama 4 Maverick at 110 tokens/second on 1 H100, with 19 MB proofs.
Read the full story on MarkTechPost.
Experts weigh in on the promises and pitfalls of Google’s new policy, which leaves schools unable to use Search unless they opt into its AI tools.
As part of the company’s post-midterms marketing campaign, data centers could be called “AI compute campuses.”
Graduates are spending more time managing AI, but are they missing out on other work experience?
MacPaw, the makers of CleanMyMac and Setapp, have an AI assistant called Eney that’s designed for Apple users. Eney is available across Mac and iPhone and is built to integrate with core Apple services like Mail, Calendar, and Reminders, plus a lot more. more…
People, including on this staff, keep asking me with varying degrees of excitement or trepidation when the AI bubble will burst.
NVIDIA shows how 3 stacked optimizations speed up MoE inference on a single DGX Spark. On Ant Group inclusionAI’s 124B Ling-3.0-flash, GSM8K decode throughput rose 42.5% with no detectable accuracy change.
Soniox has released Soniox STT v6, the latest generation of its speech-to-text API. Soniox STT v6 is a speech-to-text model that reads what a voice agent says while hearing the caller. It uses that text to recognize short replies and to drop the agent’s own voice.
OpenAI built Pando, a Git-compatible backend for its monorepo. It handles branches independently, reuses stored packfiles across fetches, cut median push latency by nearly 80%, and now delivers nearly a petabyte of Git data per day.
Zyphra Hierarchical Attention (ZHA) is a tree-indexed sparse attention method that retrofits frozen LLMs with 4B tokens of distillation, matching dense attention on reasoning and 256K retrieval benchmarks.
69 active sources
24 Turkish · 45 English
All sources
