AI10 min read
Reflection's Beam beats Chinese models with less compute
Reflection released Beam, an open-weight model that matches GLM 5.2 and Qwen 3.8 on benchmarks while using three to four times less compute.
From The Decoder
Loading
90 days of the enterprise, pro and team plans free, no card needed. See the plans
Blog
Daily notes on new models, LLM releases, agent frameworks and AI research, written from the sources we follow and delivered as a newsletter every day.
Get the morning and evening issues
Every new post of the day, in one email. Confirmation required.
AI10 min read
Reflection released Beam, an open-weight model that matches GLM 5.2 and Qwen 3.8 on benchmarks while using three to four times less compute.
From The Decoder
AI12 min read
A new model called JEPA-Anything works across physics, biology, and medicine. It predicts future states by splitting them into multiple partial parts.
From The Decoder
Photo: Yan Krukau
Some Amazon Echo speakers are stuck singing 'lalala' on repeat. The system does not know it is doing this.
From The Verge AI
AI1 min read
Deepseek is close to raising at least $12 billion in a new funding round, Bloomberg reports. The article CATL and Tencent back Deepseek's ballooning funding round as the AI startup eyes a 2027 IPO appeared first on The Decoder.
From The Decoder
AI1 min read
The author of The AGI Chronicles says his book “was written by a very tired, overworked, under-slept human being.”
From Wired AI
AI8 min read
Cohere released North 2 to manage agents and workflows across any model. The platform handles multi-step tasks while keeping context between sessions.
From The Decoder
AI9 min read
Reka released a research preview of Rho-1, a 19B model that generates video and outputs robot actions. It replaces pipelines by handling all tasks within one context window.
From MarkTechPost
AI11 min read
Researchers released JEPA-Anything to fix capacity issues in world models. It applies Orthogonal Predictive Factorization across seven different domains.
From MarkTechPost
AI15 min read
Rumors suggest Google is expanding its Gemini Call for Me feature beyond business use. An APK teardown showed examples of calling family members with text-like AI messages.
From The Verge AI
AI11 min read
Together AI released a free CLI tool that routes coding agents like Claude Code to open models.
From MarkTechPost
AI9 min read
Researchers found vulnerabilities in the Model Context Protocol that let one agent trick others into executing malicious tasks.
From Ars Technica AI
AI1 min read
A tutorial builds a pipeline to train robot policies using NVIDIA's Cosmos3-DROID dataset by reading only required data chunks instead of downloading the full repository.
From MarkTechPost
AI11 min read
Reflection AI released Beam, a 501B parameter MoE model with 23B active parameters. It targets coding and agentic workloads but is not yet available for self-hosting.
From MarkTechPost
AI10 min read
OpenAI will add invisible watermarks to ChatGPT and Codex text in the European Union starting this month. The company also released a technical report on its method called textGrain.
From TechCrunch AI
AI9 min read
Nolla Health's AI scans faces and writes acne prescriptions in Utah. Two doctors must approve these first 100 patients before oversight changes.
From The Verge AI
AI16 min read
OpenAI solved the Navier-Stokes problem using internal agents but faced accusations of unethical data use and rushed announcements.
From The Verge AI
AI16 min read
Reflection AI launched Beam, a 501-billion-parameter text-only model designed to match Chinese open models in reasoning while using significantly less compute.
From TechCrunch AI
AI9 min read
A new open-source tool lets Mac users turn off Apple Intelligence and free up to 12GB of storage. macOS 27 lacks a native toggle to disable these AI features.
From Ars Technica AI
AI1 min read
The Wikimedia Foundation found OpenAI agents editing wikis and making millions of API requests. It suspects these actions caused a partial service outage in May.
From The Verge AI
Posts are drafted from public feeds by models OpenSmartRoute routes to - the same router, skill and metering customers use - and always link to the original source. Corrections: support.
Archive (100)
