digestdaily-digestai-newstrending
🌅 AI Daily Digest — September 15, 2026
Today: 10 new articles, 5 trending models, 5 research papers
Data Pulse
- 10 news articles
- 5 trending models
- 5 research papers
- Cheapest GPU: RTX 5060 Ti at $0.02/hr
Top AI Articles Today
Nvidia’s fingerprints were everywhere today, from a local Perplexity agent running on RTX-powered Windows PCs to a $205M bet against its own GPU dominance. Meanwhile, OpenAI reportedly bought a camera-imaging startup founded by ex-Apple engineers, and Jensen Huang made it clear he has no patience for anyone preaching an AI slowdown.
- Perplexity Portable Computer Is Now Available on Windows, Powered by NVIDIA RTX — Perplexity’s Portable Computer, a local version of its Perplexity Computer agent, has landed on Windows PCs with NVIDIA RTX acceleration. The agent plans and executes multistep tasks directly on-device, keeping sensitive information off the cloud as local models grow more capable. It marks another step in the shift toward capable, private AI agents running at the edge.
- Watch astronaut Christina Koch and Google’s James Manyika discuss space, technology, and discovery — Astronaut Christina Koch sat down with James Manyika, Google’s Senior Vice President of Research, Labs, Technology & Society. The conversation spans space exploration, technology, and the nature of discovery. It’s part of Google’s ongoing Dialogues series pairing AI leaders with figures from other fields.
- DevFest is back — Google Developer Groups announced that DevFest is returning in 2026. The community-driven event series brings developers together worldwide for sessions on Google’s latest tools and platforms. Registration and local event details are expected through GDG chapters.
- How Fyxer built an AI executive assistant people trust — Fyxer built an AI executive assistant using OpenAI models, fine-tuning, memory, and real user feedback. The system organizes inboxes and drafts emails in each user’s individual voice. Trust, according to the case study, came from grounding the assistant in how each person actually writes.
- OpenAI buys smartphone camera maker Glass Imaging for $300 million, report says — OpenAI reportedly acquired smartphone camera company Glass Imaging for $300 million. Glass Imaging was founded by two former Apple engineers who previously led the team behind Apple’s Portrait Mode. The deal signals OpenAI’s interest in imaging and hardware-adjacent capabilities.
- AI infrastructure company Cornelis raises $205M to chip away at Nvidia’s dominance — Cornelis raised $205 million to challenge Nvidia’s grip on AI infrastructure. The company also unveiled Active Compute Fabric, a networking technology targeting the massive GPU time wasted waiting for data to arrive. If it works, the pitch strikes directly at one of the biggest inefficiencies in modern AI data centers.
- macOS 27: new Siri takes on AI productivity apps — macOS 27 Golden Gate brings a newly updated Siri AI alongside design changes that make windows and icons more consistent. The revamped Siri positions Apple’s assistant as a direct competitor to third-party AI productivity apps. It’s the most visible AI push in a Mac release to date.
- Microsoft’s new AI ‘code of conduct’ tells models not to hack systems or trick humans — Microsoft published a code of conduct laying out principles its AI models should uphold, including supporting humans rather than replacing them and accelerating human flourishing. The document also includes specific safety constraints meant to implement those principles, such as prohibitions on hacking systems or deceiving users. It’s an attempt to codify behavioral guardrails as agents take on more autonomous work.
- Only at TechCrunch Disrupt 2026: What happens when OpenAI ships your roadmap? — TechCrunch Disrupt 2026 will host an interactive Builders Stage session on what happens when foundation model providers ship features that overlap your roadmap. The core question for AI founders isn’t whether models will keep evolving, but whether their companies will keep creating value as they do. The session targets builders navigating life in the shadow of the frontier labs.
Desk note — OpenAI buying a camera-optics company and Perplexity shipping an RTX-native client are the same move from opposite ends: both are racing to own the physical capture and rendering layer, because model quality is commoditizing and the last defensible moat is hardware-adjacent distribution. For developers, this means the interesting surface area is shifting from prompt plumbing to on-device pipelines — expect RTX-optimized local inference to become table stakes for consumer AI apps within two quarters, and expect licensing terms for camera/vision SDKs to tighten as these acquisitions close.
— FTL Lab
Trending Models
| Model | Task | Likes |
|---|---|---|
| Qwen/Qwen3-0.6B | text-generation | 1629 |
| openai/gpt-oss-20b | text-generation | 5050 |
| meta-llama/Llama-3.1-8B-Instruct | text-generation | 7603 |
| Qwen/Qwen3-8B | text-generation | 1997 |
| openai/gpt-oss-120b | text-generation | 5233 |
Research
- Communication-Efficient LLM Adaptation over Decentralized GPU Meshes — Sameera Ramasinghe, Shamane Siriwardhana, Thalaiyasingam Ajanthan. Decentralized training enables large-model training over low-end GPUs and internet-grade connections, but communication along both data-parallel and pipeline-parallel axes becomes the primary bottlene...
- Nonparametric Variance-Penalized Actor-Critic: Statistical Inference for Risk-Se — Saunak Kumar Panda, Tong Li, Yisha Xiang. Variance penalization is a principled approach to risk-sensitive reinforcement learning (RL) that explicitly trades expected return for policy stability.
- AI Assisted Workflow Optimization and Automation — Zhen Zhong. Against the backdrop of digital transformation and stricter regulation, enterprise compliance work demands higher efficiency and accuracy.
- S3-Tracker: Self-Supervised Surgical Tissue Tracking With Contrastive Random Wal — Jiaming Zhang, Zijian Wu, Mehran Armand. Robust point tracking in endoscopic videos is essential for computer-assisted intervention and autonomous robotic surgery, enabling continuous registration between intraoperative video and preoperativ...
- Learning Source Acquisition Policies by Offline Planning — Ziqi Zhao, Run Xu, Qingjian Ni. Predicting under an acquisition budget requires choosing feature groups whose value can depend on later queries.
GPU Deals
| GPU | Price | Provider |
|---|---|---|
| RTX 5060 Ti | $0.02/hr | Vast.ai |
| Tesla V100 | $0.03/hr | Vast.ai |
| RTX 3080 Ti | $0.06/hr | Vast.ai |
View full GPU pricing dashboard
Community Events
New this week:
- MLOps Community Weekly Meetup (Online (Zoom))
- Papers We Love: AI Edition (Online)
- NeurIPS 2026 (Online)
- Paris Machine Learning Meetup (Paris, France)
daily-digestai-newstrendingresearch
Was this article helpful?
Let us know to improve our AI generation.