AI Briefing
TLDR AI news
The latest AI and developer news from TLDR AI, with short summaries.
Visit website →NVIDIA Spectrum-X sets gigascale AI standard with open AI-native Ethernet fabric, now supporting MRC
NVIDIA Spectrum-X has put forward a gigascale AI networking standard with MRC and multi-plane support.
Apple Considers Introducing Multiple AI Models in iOS 27
Apple is expected to let users choose from multiple AI models in iOS 27.
OpenAI Releases Separate Enterprise ChatGPT iOS App
OpenAI has separately released a ChatGPT iOS app for Microsoft Intune.
Alphabet rises on report of Anthropic's $200 billion cloud spending over the next 5 years
Alphabet rose on a report that Anthropic will spend $200 billion over 5 years on cloud.
Google Prepares Gemini Flash Model Upgrade
Google is preparing the next upgrade to Gemini Flash around I/O 2026.
AI2 unveils MolmoAct 2
AI2 released MolmoAct 2, an open foundation model for real-world robots, along with a dataset of over 720 hours.
Gemini API File Search Now Supports Multimodal RAG
Gemini API File Search now supports image-text RAG and page citations.
Google revisits the hallucination problem through the lens of uncertainty
It argues that reducing hallucination requires metacognition and faithful expression of uncertainty.
How to Scale Models
A summary of the principles and parallelization strategies for scaling LLM throughput near-linearly on TPUs and GPUs.
Finding the Wasted Bits: How Much Information Do LLM Weights Actually Carry
Analysis shows LLM weights in BF16 carry only about 10.6 bits of actual information out of 16 bits.
Subquadratic Breaks the Context Window Limit, Unveils 12 Million Token Window
Subquadratic touted a 12 million token context window with SSA and an 83 score on MRCR v2, surpassing GPT-5.5.
Accelerating Gemma 4: Faster Inference with Multi-Token Prediction Drafters
Google has released MTP drafters for Gemma 4, boosting inference speed by up to 3x.
Computer Use Is 45x More Expensive Than Structured APIs
AI agents that directly manipulate UIs via vision are far less efficient in both cost and reliability than those using structured APIs.
GPT-5.5 Instant: Smarter, Clearer, and More Personalized Answers
OpenAI has updated ChatGPT's default model to GPT-5.5 Instant, improving accuracy and personalization.
Google Partners with XPRIZE and Range Media Partners on $3.5 Million Future Vision Film Competition
Google is supporting the $3.5 million Future Vision XPRIZE film competition with XPRIZE.
Meta Plans to Develop Advanced 'Agentic' AI Assistant for Users
Meta plans to launch a personalized AI agent that performs everyday tasks based on the new **Muse Spark AI** model before Q4 of this year.
White House Reviews Pre-Release Screening of AI Models
The White House is considering a government pre-screening system before AI models are released to the public.
How LLMs Distort Our Writing
LLMs don't just fix grammar—they distort meaning by shifting the argument and tone of writing.
End-to-End Tokenizer Training for Autoregressive Image Generation
By jointly training the tokenizer and the generative model, the approach achieved FID 1.48 on ImageNet 256x256.
Inside DigitalOcean's AI-Native Cloud Leading the Inference Era (7 min read)
DigitalOcean unveiled an AI-native cloud built for inference and agent workloads.