AI Titus's Morning Wire

AI NEWS TODAY

· Edited by Titus Blair
META JUMPS INTO THE PAID-AI RACE: MUSE SPARK 1.1 SHIPS AS A MILLION-TOKEN CODING MODEL, AND ZUCKERBERG BREAKS A 3-YEAR X SILENCE TO SELL IT
★ Must-Read / Watch agentic engineering / build-better-agents📚 Learn state of AI coding / useful technique

⚡ AI Frontier · smol.ai

★ Must-ReadAnthropic launches Reflect, a Claude usage dashboard that nudges you to log off
A new beta dashboard shows how you have used Claude over the past months, like a 'Claude Wrapped,' with reflection prompts and optional break reminders. It excludes health and incognito chats, and Anthropic pitches it as healthier use rather than more use.
★ Must-ReadMeta's paid Muse Spark API undercuts rivals by roughly 4x
The practical companion to the lead: Muse Spark 1.1 is Meta's first paid model API, at $1.25 in / $4.25 out per million tokens with $20 in free credits, about a quarter of what comparable OpenAI and Anthropic models cost.
📚 LearnMicrosoft 365 Copilot's July drop makes Claude selectable inside Office
Microsoft's monthly Copilot update lets business users pick Anthropic's Claude in Copilot Chat, moves Copilot Cowork to general availability, and adds watermarking for AI-generated content across the Office apps.
Accenture and Google Cloud ship prebuilt agents aimed at mid-market firms
A joint suite of ready-made agentic AI for companies with $300M to $3B in revenue, a sign the agent land-grab is moving down from the giants to mid-size businesses.

📺 Watch · latest videos

★ Must-WatchMeta Is Back: First Thoughts on Muse Spark 1.1
A hands-on first look at Meta's new paid coding model, walking through what it does well and how it feels against the current leaders.
Fahd Mirza
Meta Just Dropped Muse Spark 1.1 and the AI Race Has a New Leader
A rundown of the launch, Zuckerberg's return to X to announce it, and how the million-token model stacks up on price and capability.
AIM Network
OpenAI's Double Bet: Smarter Models, Autonomous Work
A daily-news episode covering the week's frontier moves across OpenAI, Meta, and the enterprise-agent push.
GAI Insights
📚 LearnExciting AI Updates Weekly - July 10, 2026
A calm weekly roundup of the launches, papers, and tools from the week, aimed at practitioners keeping up.
Lev Selector
AI News in a Minute - July 9, 2026
A one-minute skim of the day's headlines for people who just want the gist before moving on.
Jerome W. Dewald

🗣 Voices & Blogs

★ Must-ReadMuse Spark 1.1: everything you need to know
Artificial Analysis benchmarks Meta's new model independently: it jumps 8 Intelligence Index points in three months, and its hallucination rate falls 35 points mostly by learning to abstain rather than guess.
📚 LearnThe July 2026 Frontier: Muse Spark 1.1, Grok 4.5, GPT-5.6 Sol
Ken Huang steps back from the week's back-to-back launches to compare where each of the four big labs actually leads, and what it means for picking a model for real work.
I tested Meta Muse Spark against 4 frontier models
An independent hands-on run pitting Muse Spark 1.1 against the current top models on real coding and reasoning tasks, rather than repeating the launch benchmarks.
Meta on the offense
A business-strategy read on why Meta is abandoning its free-and-open posture to charge for Muse Spark now, and what the shift says about the economics of frontier AI.
Microsoft joins Google in backing Go for AI agents — OpenAI and Anthropic lag
Go has emerged as the lingua franca for cloud infrastructure, used for everything from container orchestration and CI/CD pipelines to The post… · The New Stack

🌏 The Wire · Drudge / Breitbart

NYT and other publishers ask a judge to sanction OpenAI over hidden evidence
The plaintiffs say a deposition revealed OpenAI had searched its training data for copyrighted news, and kept 78M de-identified ChatGPT chats to gauge infringement, after telling the court that was infeasible. OpenAI calls the claims false.
China forces Doubao and Qwen to kill custom AI agents before July 15
New 'anthropomorphic AI' rules require anti-addiction limits and instant-exit that persistent-memory agents cannot meet, so ByteDance and Alibaba are switching off user-created agents; Qwen will permanently delete the configurations.
China also weighs curbing global access to its top Alibaba and ByteDance models
A companion policy push would tighten who outside China can use the country's frontier models, even as those same open models keep gaining share with US developers on cost.
Agility Robotics to go public via SPAC, the first pure-play humanoid on the market
The maker of the Digit warehouse robot will list through Churchill Capital, though its CEO is careful to tamp down expectations of a robot in your home any time soon.
Ollama raises $65M as its open-model runner hits 8.9M developers
The tool that lets you run open models locally with one command took a Series B led by Theory Ventures ($88M total raised), and says it is now used inside 85% of the Fortune 500, an open-model counterweight to the week's paid launches.

📈 Markets

NVDA 210.96 ▲4.0%
MSFT 385.10 ▲0.2%
GOOGL 357.18 ▼0.5%
AMZN 245.34 ▼0.7%
META 669.21 ▲6.0%
AMD 557.89 ▲2.0%
AVGO 399.97 ▼0.3%
PLTR 126.79 ▼1.7%
SPCX 145.30 ▼4.5%
TSLA 407.76 ▲0.3%

The BriefMeta released Muse Spark 1.1, an AI model built to write and fix code and operate software on its own, and for the first time Meta is charging developers to use it through a paid API instead of giving its AI away for free. It can hold about a million tokens of context at once (roughly a small book's worth of code and instructions) and Meta says it matches or beats the current top models from Anthropic and OpenAI on real coding work. It matters because Meta spent years as the big free, open alternative, so putting its best model behind a paywall is a real strategy shift, and it drops a fourth deep-pocketed player into a fight that already has OpenAI, Anthropic, and SpaceXAI.

Level UpAs AI coding models get better at running on their own, the durable skill is shifting from writing code to designing the loop the agent runs in: what it should attempt, how it checks its own work, and when it should stop. Armin Ronacher, the creator of Flask, argues exactly this in 'The Coming Loop.' Try it this week: take one repetitive task you normally do by hand and sketch it as an agent loop with a clear success check, instead of just prompting once. The Coming Loop (Armin Ronacher)