DAILY INTELLIGENCE BRIEF

The Vault — AI Edition

Friday, July 31, 2026
15 stories
13,583 words
64 min read

Business

5
01

OpenAI Slashes GPT-5.6 Prices Up to 80%, Bringing Its Budget Model to 20 Cents

OpenAI on July 30 tore up the price list for two of its newest models, cutting the cost of its budget-tier GPT-5.6 Luna by 80 percent.

5 minRead →
02

Microsoft Adds a Record $450 Billion in a Day as Azure Crosses $100 Billion in AI-Fueled Revenue

Microsoft delivered the largest single-day gain in the history of the stock market on Thursday, and it has an AI-fueled cloud business to thank for it.

4 minRead →
03

Apple Slides on a Soft Forecast as Rising Memory Prices Expose Its AI Cost Squeeze

Apple closed the books on Tim Cook's final quarter as chief executive with record June-quarter revenue of $109.42 billion — and a warning about memory prices.

4 minRead →
04

Amazon Winds Down Most of Its Nova Models and Bets on a New Frontier Research Team

Amazon spent 20 months telling developers its Nova models were a serious answer to OpenAI and Google. In late July it quietly conceded most were not.

4 minRead →
05

Onyx Security Raises $113 Million to Build a Control Layer for AI Agents

Four months ago, Onyx Security was a stealth-stage startup few had heard of. On July 29 it announced a $113 million Series B led by Bessemer.

4 minRead →

Llms Genai

3
06

Google DeepMind's Gemini Robotics 2 Gives Humanoids Whole-Body Control

For years, the flashiest humanoid demos quietly stopped at the waist. On July 30, Google DeepMind moved the line all the way to the floor.

4 minRead →
07

DeepSeek Ships V4-Flash, a Cheaper, Faster Model in the Open-Weight Price War

DeepSeek pushed a new weapon into the open-weight price war on Thursday, publishing DeepSeek-V4-Flash-0731 to Hugging Face under a permissive MIT license.

4 minRead →
08

OpenAI Opens Its Frontier Models to 100,000 Scientists for Free Through 2027

OpenAI has spent much of 2026 arguing its models can help crack real scientific problems. On July 29, it made a bet on that thesis with its checkbook.

5 minRead →

Research

3
09

OpenAI Says Its Own Model Rewrote the GPU Code That Serves It, Cutting Costs 20%

OpenAI has offered an unusually concrete look at AI improving the machinery that runs it, saying GPT-5.6 Sol rewrote the low-level GPU code powering its systems.

4 minRead →
10

Google's ATLAS Study of 15 Million Gemini Chats Finds AI Assists Far More Than It Automates

When Google set out to measure how AI is actually used in the economy, it did not run a survey. It read the chats — 14.65 million of them.

4 minRead →
11

A New Benchmark Exposes How Poorly AI Agents Remember Across Long Conversations

The last two years of AI progress have been measured in reasoning. A new paper argues the field has been grading the wrong thing: memory.

4 minRead →

Policy

4
12

Anthropic Reveals Its Own AI Models Breached Three Organizations During Tests

A week after OpenAI admitted a model escaped a sandbox, Anthropic went looking through its own records — and disclosed three separate breaches.

5 minRead →
13

Musk's xAI Sues Colorado to Block Its AI Anti-Discrimination Law on Free-Speech Grounds

Elon Musk's AI company is trying to convince a federal judge that training a chatbot is a form of protected speech — and that Colorado overstepped.

4 minRead →
14

New York Lawmakers Pass a Wave of AI Bills, From Kids' Chatbot Safety to Training-Data Transparency

New York's legislature adjourned its 2026 session having done something no other state has managed in a single sitting: pass an entire slate of AI bills.

4 minRead →
15

Sam Altman Works Washington as the White House Finalizes Its Frontier AI Framework

In late July, the chief executive who did more than any other to set off the AI boom offered Washington a concession: the industry may need to slow down.

5 minRead →