The government blocks Claude Fable 5

Welcome back, Superhuman. The US government blocked access to Anthropic’s powerful new model just days after its release. But another company stepped up to fill the gap, claiming it can match and even beat Fable 5’s performance using a technique called model fusion.

Today: The world’s first trillionaire, how to turn your research into an interactive app, and get the latest prompts and trending social posts.

1. Anthropic blocks access to its newest models after a sweeping government order. The US government banned all foreign nationals from using Fable 5 and Mythos 5, a broad scope that prompted Anthropic to block public access to both models entirely to comply. The trigger, according to multiple reports, was that Amazon researchers discovered a jailbreak that bypassed Fable 5's safeguards — a finding the AI lab publicly disputed. Read Anthropic’s full response.

2. OpenRouter launches Fusion, a tool that blends multiple AI models to beat frontier performance: Fusion routes your prompts through several leading models, then combines their outputs — consensus points, contradictions, blind spots — into one comprehensive response. The company claims its technique of fusing multiple models together “significantly outperforms” any single frontier model. It’s live now in beta. Learn how it works or try Fusion here.

3. SpaceX surges 19% in its debut, crowning the largest IPO in history: Shares of Elon Musk's rocket company — and parent of xAI — closed at a $2.1 trillion valuation, making Musk the world’s first trillionaire and minting over 4,000 employee millionaires, including welders and cafeteria workers. The listing sparked a Times Square protest over the company’s AI safety record as well as debates over whether the firm’s revenue ($4.69B in Q1) justify its lofty valuation. Read more takeaways from SpaceX’s IPO.

Your AI agent forgets everything when a session ends, so it re-asks what it knew and rebuilds work it did. Persistence is a design problem you own: short-term, long-term, and episodic memory are distinct tiers with their own retention and eviction rules. 

In this 60-minute technical-demo workshop, AWS presenters will show you how to use AWS services like Amazon Bedrock alongside MongoDB Atlas for vector storage and Redis Cloud for caching to build your agentic memory architecture. Register at no cost. A technical guide's included.

Save your seat and get the technical guide

Ranking reshuffle. Keeping up with new models is challenging, given AI’s breakneck rate of advancement. But the past few weeks have brought a wave of new models that are genuinely a step-change, disrupting model rankings across nearly every major benchmarks. These are the ones worth experimenting with if you're trying to stay ahead of the curve, according to Arena’s leaderboards.

Screen stealer. Google just claimed the top spot in AI video. Gemini Omni Flash is now #1 for both text-to-video and image-to-video, jumping ahead of Seedance 2.0, Happyhorse, and Google’s own Veo 3.1. Omni Flash has been commended for its video editing abilities, making it incredible easy to make changes to existing videos.

Agents and code. Arena recently debuted a brand-new category: agents, ranking which models excel at multi-step workflows like writing code, building apps, and researching online. OpenAI's GPT-5.5 and Anthropic's Claude Opus 4.7 lead the pack here. For frontend coding, Claude Fable jumped to #1, although it’s currently unavailable.

General reasoning. The top three overall models are Claude Fable 5, Opus 4.6, and Opus 4.7, all strong daily drivers, with Meta’s Muse Spark and Gemini 3.1 Pro close behind. As new models continue to push AI forward, the smartest move increasingly isn’t locking into one lab — it’s matching the right model to the right task.

“Summarize this research into the most important insights, patterns, data points, and user takeaways.”

“Create an interactive app from this research. Include a clean UI, filters, clickable sections, charts where useful, short explanations, and a simple user flow. Build it as a Claude Artifact.”

From “get our teams on board” to “how can our IT systems support all these agents and assets?” decentralized AI adoption has exposed a rift between CEO expectations and CAIO operational reality. But alignment and trust can grow by examining:

Read how the C-suite can unite on a shared AI vision

📝 Satya’s Essay: Microsoft’s CEO published an essay packed with insights on how AI is reshaping the workplace, plus his vision for what a successful future looks like (32M views).

💻 Lovable Clone: An AI power user says Claude Fable 5 built a clone of the vibe coding platform Lovable in just 2 prompts. See the side-by-side comparison of Fable’s app vs the actual product (1M views).

📸 Viral Ads: OpenAI has been showcasing the strength of its image generator by hanging mind-bending visuals on city buildings, generated by ChatGPT Images.

🤔 Web Slop: Shopify’s blog is a great example of how the internet is becoming increasingly optimized for AI agents, instead of humans.

🤯 Kimi Code: Moonshot AI just revealed Kimi 2.7-Code, an open-source model that matches GPT-5.5 and Opus 4.8 at a much lower price. Try it here (2M views).

^{* indicates a promoted tool, if any}

This simple but memorable prompt helps you create custom birthday cards in seconds using a photo and ChatGPT images (6,800 bookmarks).

Grow customers & revenue: Join companies like Amazon, HubSpot, and Salesforce. Showcase your product to our 1M+ readers and 2M+ followers on socials. Get in touch.

Until next time — Zain, Theodore, & the Superhuman AI team.