SIGNAL

Tuesday, May 26, 2026

6 stories

snapshot

→ signal --date=2026-05-26

Today's AI landscape is defined by three converging forces: DeepSeek's permanent 75% price cut is compressing margins industry-wide, leaked code suggests Anthropic's powerful Mythos model is launching sooner than expected, and the Middle East conflict is forcing a reckoning with the physical vulnerabilities of AI infrastructure. Meanwhile, Claude Code's move to the web and tiny models like Cactus Needle show the technology branching in opposite directions — toward cloud ubiquity and edge independence simultaneously.

→

DeepSeek locks in a permanent 75% price cut as the AI price war escalates, leaked code suggests Claude Mythos is launching sooner than expected, and the Middle East conflict puts Gulf AI infrastructure ambitions to the test.

AI Build Fast with AI

DeepSeek Makes 75% Price Cut Permanent — AI Price War Enters Most Aggressive Phase

DeepSeek announced its 75% temporary discount on V4-Pro is now permanent, making frontier-class Chinese reasoning available at roughly one-quarter the cost of equivalent OpenAI or Anthropic tokens. Chinese models already represent 60% of OpenRouter usage.

AI Build Fast with AI

Leaked Code Suggests Claude Mythos Launch Is Imminent

Code preparations for Claude Mythos 1 surfaced on GitHub between May 22-24, including reasoning protocols and production references to an Opus 4.7 thinking variant, strongly suggesting Anthropic's next major model launch is weeks away.

MENA CNBC

Middle East Conflict Puts Gulf AI Infrastructure Ambitions to the Test

Attacks on data centers in the UAE, closure of the Strait of Hormuz, and oil at $100/barrel are forcing AI infrastructure investors to reassess the Gulf's viability as a global AI hub. Some investment decisions have been paused.

DevTools Tom's Guide

Claude Code Ships to the Web — Agentic Coding Goes Device-Free

Anthropic launched Claude Code on the web and iPhone, moving the agentic coding tool beyond the terminal for the first time. Developers can now run long coding sessions from any browser without keeping a laptop open.

Business Build Fast with AI / Dentro

Microsoft Raises 2026 AI CapEx to $190 Billion as Infrastructure Costs Surge

Microsoft now expects 2026 capital expenditure to hit $190 billion, with $25 billion of the increase attributed to surging memory and storage component prices driven by AI infrastructure demand. The company has spent $97 billion in the last four quarters.

DevTools Radical Data Science

Cactus Needle: 26M-Parameter Model Runs Locally at 6,000 Tokens/Second

A 26-million-parameter model distilled from Gemini 3.1 achieves 6,000 tokens/second prefill and 1,200 tokens/second decode on consumer hardware. The fully open weights point to a future of capable on-device AI for phones, watches, and glasses.