Good morning {{first_name| Artificial intelligence lovers}}. The frontiers of AI will not remain stable for long, and this week, the pendulum is swinging back toward OpenAI.
The company’s new GPT 5.5 ‘Spud’ model capped off a big week of releases with a jump in the leaderboards and a strong shift in the air… landing at the precise moment Anthropic is absorbing its harshest week of maximum price and quality complaints in months.
Note: We love to hear how our readers are applying AI in their daily lives. Share your community’s workflow here (With a new audio submission and interview streaming option) for a chance to be featured in an upcoming newsletter.
-
OpenAI restores boundaries with GPT 5.5
-
The United States warns of the theft of artificial intelligence in Chinese laboratories on an industrial scale
-
Get a news brief every morning with Claude
-
The biggest AI productivity winners are also the most worrisome
-
4 new tools for AI, community workflows, and more
Obinay
Rundown: OpenAI only Fired GPT-5.5 (codenamed “Spud”), is the company’s long-awaited upgrade, pitched as a “new class of intelligence” – topping benchmark scores across the industry and beating Anthropic on the frontiers of the AI paradigm.
-
5.5 The highest levels are determined through a series of tests of reasoning, computational use and coding of general models, with several grades comparable By Claude Mythos.
-
The model maintains the same speed as 5.4 with added efficiency, as OAI says it used Codex and 5.5 to rewrite its GPU code to improve the infrastructure.
-
GPT-5.5 is priced at $5/$30 per million I/O tokens for API pricing, with OAI viewing it as “half the cost of competitive frontier tokenization models.”
-
Version 5.5 is rolling out across ChatGPT plans and in Codex with Thinking and Pro variants, with OAI continuing to highlight the “generous usage” of its new releases.
Why it matters: After months of human dominance, the mood is changing again – with OpenAI quickly shipping powerful new upgrades and reviving magic that seemed a bit lost in previous releases. With Anthropic now wading through complaints of rate caps and deteriorating quality, this is a big week for SMA and its partners. On the feelings front.
Along with ORIX
Rundown: Agentspan is an open source framework and runtime designed to build, run, and monitor persistent agents. Join the upcoming webinar to see how modern engineering teams leverage Agentspan to build resilient agents that hold up in the real world.
The April 30 session will cover:
-
The 4-layer production stack every AI agent needs to achieve robustness at scale
-
How to make existing frameworks robust, including LangGraph, OpenAI Agents SDK, and Google ADK, using Agentspan
-
Real-world patterns to keep agents alive when operations fail
Artificial intelligence and geopolitics
Image source: Images 2.0 / Rundown
Rundown: The white house published A memo accuses Chinese companies of “industrial-scale” distillation campaigns against frontier AI labs in the US – and comes weeks before a scheduled Beijing summit between Trump and Xi Jinping.
-
Distillation trains smaller AI systems on the output of frontier models that Kratsios says China runs through thousands of fake API accounts and jailbreaks.
-
Anthropic accused DeepSeek, Moonshot, and MiniMax of distillation in February, with this memo raising the particular complaint to federal policy.
-
The Chinese embassy denied these accusations, describing them as “pure slander,” ahead of Trump and Xi’s meeting in Beijing on May 14-15.
-
A House-passed foreign affairs bill that received its first vote this week would prompt the administration to add distillation violators to a U.S. export blacklist.
Why it matters: Dario Amodei recently framed open source and China as frontier laboratories 6 to 12 months behind. Cratesius’ memorandum reframes this gap, arguing that gains come from scrapping tactics, not from acts of architecture. Whether this is true depends on how much of the DeepSeek/Kimi trail is really tracking distillation versus new research.
Artificial intelligence training
Rundown: In this guide you’ll learn how to turn your daily updates into a personalized newspaper with Claude. It’s useful because it gives you a single-label summary of important stories, action items, and calendar settings.
-
In Claude or Claude Cowork, make sure Slack, Notion, Gmail, and Calendar are connected
-
Prompt: “Create a consistent morning edition of my Slack, Notion, Gmail, and Calendar updates from the past 24 hours. Categorize what’s most important and format it like a newspaper with top stories, action items, and onboarding schedule”
-
Review the first draft and give feedback. Tighten order, planning and confirmation. Then ask Claude to turn your workflow into a skill
-
In Claude Cowork, create a recurring task that turns this skill on each morning
Pro Tip: Have other agents collect external news and drop it into the Notion database. Then use this skill as an editor to turn everything into a daily newspaper.
Provided by OPTIMIZELY
Rundown: It’s time to use AI in your marketing; Embedded into your workflow and eliminating tedious work. Consider this your official invitation to Opal U: AI Marketing University from the Optimizely Opal team.
Here’s what you’ll get:
-
Live workshop (free) with 50 top marketers – 1 hour daily (Mon-Fri)
-
3 working AI agents that you build and bring back to your team
-
Invest 5 hours to save more than 10 hours every week
Artificial intelligence research
Rundown: Anthropic published And focused economic follow-up Cloud survey of 81 thousand usersIt found that the people who get the biggest productivity boost from AI are also the most concerned about losing their jobs to it, especially early-career workers.
-
The survey links Anthropic’s Economic Index usage data (on which Jobs relies more than Cloud) to 80,508 workers’ views on how AI will reshape their roles.
-
Workers whose jobs use Cloud are the most vocal about concerns about AI displacement, three times more than those whose jobs use less, with engineers leading the concern.
-
Most participants said AI gains through faster tasks and free time, but also leads to scope expansion and more work.
-
Early-career respondents expressed highest concerns about displacement, supporting Anthropic’s previous indication of slowing hiring of recent graduates in the US
Why it matters: The conventional view is that AI panic will come from lower-level users, but these results flip that, with anxiety coming from those who get the most out of the tools. Despite the productivity boosts, AI sentiment has never been lower – and there don’t seem to be many solutions on the horizon to ease tensions.
-
🥔 GPT 5.5 – The new top-rated leading AI model from OpenAI
-
⚙️ Super review – Claude Code command to run multi-agent code reviews
-
⚙️ Qwen3.6-27B – The new 27B model from Alibaba, which is superior to its predecessor
band Launched the missing infrastructure layer for multi-agent systems. Any agent. Any environment. See it in action.*
Anthropic published Claude Code’s post-mortem complaints for three separate bugs followed, and usage limits were reset for subscribers due to the issues.
OpenAI foot ChatGPT for Clinicians, a free tool for certified health professionals in the US, GPT-5.4 received a score of 59.0 on HealthBench Pro, topping Clinicians and Opus 4.7.
dead sender An internal memo to employees informing them that the company will lay off 10% of its workforce in May, due to AI efficiency and other investments.
Elon Musk SpaceX And it is said In conversations With French artificial intelligence startup Mistral in a three-way partnership, in addition to its recent deal with programming startup Cursor.
Tencent Open source Preview of Hy3, its first model of the reconstructed training set with competitive proxy encoding and search proxy results to the best open models.
In each newsletter we show how the reader is using AI to work smarter, save time or make life easier.
Today’s workflow comes from the reader Mark M. In Virginia:
“I used AI to turn the headache of tax season into a streamlined system. Every year, the same scene plays out: I drop off donation bags, hide the receipt, and then scramble at tax time trying to decipher my own list and determine the value of each item.
This year, I photographed handwritten donation lists and uploaded them to Claude AI. Claude transcribed my listings, asked clarifying questions, verified charities with the IRS, grouped items into logical categories, applied appropriate FMV to the case, and produced professional PDF summaries – all via conversation, in minutes.
Multiple donations and nearly 400 items in fully documented and defensible deductions. “For anyone who donates, this is the workflow you didn’t know you needed.”
How do you use artificial intelligence? Tell us here.
That’s all for today!
Before you go, we’d love to know what you thought of today’s newsletter to help us improve The Rundown experience for you.
Rowan, Joey, Zack, Shobham and Jennifer – the humans behind The Rundown
