The AI Daily Brief: "All of AI's New Models and Tools" – April 9, 2026

Host: Nathaniel Whittemore (NLW)

Episode Overview

This episode provides an in-depth analysis of the latest developments in the world of artificial intelligence, focusing on groundbreaking model releases, major enterprise tools, and the broader implications for industry and society. NLW breaks down headline stories—like OpenAI's new model rollout controversy and Anthropic’s ongoing legal struggles—before dedicating the main segment to the week's most important AI model and tool launches from Meta, Z AI, Anthropic, and Google.

Key Discussion Points and Insights

1. OpenAI's Spud Model Controversy

Timestamps: [02:00] – [06:20]

• Initial Headline: Reports emerged via Axios that OpenAI planned a staggered, limited release of a new model named "Spud" due to cybersecurity risks, echoing Anthropic's earlier "too powerful to release" narrative.
• Twitter/X Reaction:
  • Daniel Mack: "Breaking OpenAI will not release Spud. Dario forced their hand. Total anthropic victory."
  • Dan Shipper: "The new status symbol is making a model so powerful you can't release it?"
• Correction: (Shortly afterward) Dan Shipper clarifies that the Axios story conflated OpenAI’s cybersecurity product test group with Spud itself; Spud wasn’t being withheld, the story had been inaccurate.

  “My friends, we are playing with live ammunition here, but since I caught this in time to update, I wanted to make sure we did.” (NLW, [06:20])

2. Perplexity Computer: A Meteoric Rise

Timestamps: [06:20] – [09:10]

• Perplexity’s new ‘Computer’ product: Launch and shift to usage-based pricing led to doubling revenues in a quarter.
• Adoption:
  • 100 million monthly active users
  • $450 million ARR
  • Enthusiastic adoption in the finance industry, per Geiger Capital and others.
• Skepticism from some who predict competition from integrated “super app” platforms (e.g., Cowork, GPT Super App).
• Impact on GitHub: Surge in AI-generated code driving record-breaking commit stats and infrastructure strain.

3. GitHub’s Agentic Coding Wave and Infrastructure Strain

Timestamps: [09:10] – [11:55]

• Context: AI agents are causing explosive growth in commits—GitHub could hit 14 billion this year.
• Challenges: Outages, quota limits, and scaling pains as the platform adapts to both human and machine contributors.

  “This hasn’t been designed with agents in mind.” – Peter Steinberger ([10:55]) “Pushing incredibly hard on more CPUs, scaling services and strengthening their core features.” – GitHub COO Kyle Daigle ([11:30])

4. Anthropic vs. The Pentagon: Legal Battle Deepens

Timestamps: [11:55] – [16:15]

• Appeals court denied Anthropic's request to suspend “supply chain risk” designation for Pentagon contracts.
• Complex legal landscape: Simultaneous lawsuits in California and D.C. courts; conflicting implications for non-Pentagon agencies vs. the Pentagon itself.
• Notable Quotes:

  “Our position has been clear from the start. Our military needs full access to Anthropic's models if its technology is integrated into our sensitive systems.” – Acting Attorney General Todd Blanche ([15:10]) “The D.C. circuit's denial will prolong ambiguities regarding whether political considerations can drive federal procurement.” – Matt Schroers ([15:45]) “Two out of the three judges ... have been very, very sympathetic to the Trump administration's aggressive claims about executive authority in the past.” – Charlie Bullock ([16:00])

• Next Steps: Rapid movement possible toward Supreme Court resolution; implications could set precedent for tech-government relations.

Main Segment: Major New Models and Tools

5. Meta’s MuseSpark: First Frontier Model from Meta Superintelligence Lab

Timestamps: [17:15] – [24:20]

• MuseSpark Summary:
  • First model from Meta’s Superintelligence Labs, led by Alexander Wang (Scale acquisition).
  • New family replaces “Llama,” aiming to shed old baggage.
  • Natively multimodal (text, vision, reasoning), supports tools & multi-agent orchestration.
• Performance:
  • Benchmark scores: Competitive, but not leading—near Opus 4.6, Gemini 3.1 Pro, GPT 5.4 for coding and reasoning.
  • Particularly strong at visual reasoning—state-of-the-art on Charvik’s Reasoning.
• Use Case Focus:
  • Mark Zuckerberg: “Musespark is a world class assistant and particularly strong in areas related to personal superintelligence like visual understanding, health, social content, shopping, games and more.” ([20:20])
  • Agentic focus—Meta shifting from just "assistant AI" to "agentic AI" able to take actions.
• Reception:
  • Ethan Malik: “Meta's Muse spark thinking is fine so far, but really doesn't match the current Big three models... it's not bad, just not the vibe level that the benchmarks might indicate.” ([22:10])
  • Francois Chollet: “The new model from Meta is already looking like a disappointment. Over optimized for public benchmark numbers at the detriment of everything else.”
  • Alexander Wang (Meta AI): "We're always open to feedback... We have been pleasantly surprised by users' feedback in areas like visual coding, writing style and reasoning queries." ([23:25])
  • Vaas: “Meta’s latest model, musespark, is actually much better than I had expected… Is it Frontier leader in any single category? No. Is it better than I expected? Yes.” ([23:40])

6. Z AI’s GLM 5.1: A New Open Source Challenger

Timestamps: [24:20] – [27:30]

• GLM 5.1 Highlights:
  • First open source model to beat leading Western models on some coding benchmarks (Suitebench Pro 58.4 vs. GPT 5.4’s 57.7).
  • Full open source release with commercial licensing (754B parameters—enormous model).
• Technical Feats:
  • Claim: Built a Linux desktop autonomously in 8 hours, completed database optimization tests with 6000+ tool calls and significant performance gains.
  • “GLM 5.1 can do 1700 [agentic steps] right now. Autonomous work time may be the most important curve after scaling laws.” – Lou, Z AI leader ([26:50])
• Meta/Narrative:
  • Trained on less powerful Huawei chips—demonstrates China’s rapid progress.
  • "Everyone’s freaking out about Claude Mythos while Zai casually open sourced a model built for eight hour autonomous execution." – Leet LLMs ([27:15])
• Skepticism over benchmarks until third-party validation, but optimism about open source momentum.

7. Anthropic’s Claude Managed Agents: Bringing Agentic AI to Scale

Timestamps: [27:30] – [33:20]

• Launch: “Claude Managed Agents” platform enables devs to build, deploy, and monitor powerful agentic systems with minimal setup.
• Key Features:
  • Agent harness: Software infrastructure wrapping the AI model for autonomous action.
  • Built-in sandboxed environment: For secure agent execution.
  • Control over permissions, monitoring, and hours-long autonomous activity.
• Industry Reaction:
  • Angela Jiang, Anthropic: “There is a notable gap between what anthropics models are capable of and what businesses are using them for. This tool is meant to close that gap.” ([28:20])
  • Eric Liu (Notion): Demoed seamless onboarding automation using Claude agents.
  • Alex Albert, Anthropic: “Managed agents eliminates all the complexity of self hosting an agent, but still allows a great degree of flexibility…” ([29:45])
• Common Patterns: Event-triggered tasks, scheduled tasks, fire-and-forget ops, and “long horizon” tasks—showcased via community experimentation.
• User Stories:
  • Jared Orkin: "You no longer need an engineer to run an overnight marketing analysis. You need one sharp operator in an afternoon, set the schedule, set the guardrails and walk away. Anthropic runs the infrastructure. You pay per session hour." ([31:05])
  • Powell Hurren: “I built my first managed agent. Surprised how easy it was. You describe what you want in plain English. The platform generates a full agent config all in YAML you can edit.” ([31:40])
• Current Limitation: Persistent memory across sessions not yet available—better for transactional tasks than continuous learning agents.
• Forecast: NLW expects widespread adoption and deeper coverage in future episodes (harness engineering deep dive teased).

8. Google Gemini Notebooks: Product Unification and Knowledge Management

Timestamps: [33:20] – [35:25]

• New Feature: Notebooks in Gemini allow users to organize resources and context, plus assign custom instruction sets for project-specific use.
• Josh Woodward (Google):

  “Most AI chatbots give you basic projects. Gemini just built you a second brain.” “You can take the resource management you’re doing in NotebookLM and put it directly in the Gemini app.”

• Significance: Shift toward consolidating project management/user knowledge within one cohesive Google experience—potentially more practically impactful than a new model release for end users.

Notable Quotes & Memorable Moments

• Dan Shipper ([03:50]): “The new status symbol is making a model so powerful you can’t release it?”
• Mark Zuckerberg (Meta) ([20:20]): “We are building products that don’t just answer your questions, but act as agents that do things for you.”
• Alexander Wang (Meta AI) ([23:25]): “We’re always open to feedback… publish those results for the community to understand.”
• Lou, Z AI ([26:50]): “Autonomous work time may be the most important curve after scaling laws. GLM 5.1 will be the first point on that curve that the open source community can verify with their own hands.”
• Angela Jiang (Anthropic) ([28:20]): “There is a notable gap between what anthropics models are capable of and what businesses are using them for. This tool is meant to close that gap.”
• Josh Woodward (Google) ([33:45]): “Gemini just built you a second brain.”
• NLW ([36:00], closing): “For those of you who are interested in going a little bit deeper in anthropic managed agents, I think I’m going to do a main episode about harness engineering soon where we’ll dig deeper into that.”

Important Timestamps

• OpenAI/Spud Model Drama: [02:00] – [06:20]
• Perplexity Computer & GitHub Trends: [06:20] – [11:55]
• Anthropic Pentagon Legal Drama: [11:55] – [16:15]
• Meta MuseSpark Deep Dive: [17:15] – [24:20]
• Z AI GLM 5.1 Launch: [24:20] – [27:30]
• Anthropic Claude Managed Agents: [27:30] – [33:20]
• Google Gemini Notebooks: [33:20] – [35:25]

Conclusion

While the week's discourse may have been dominated by “models too powerful to release,” the rest of the AI world pressed forward with rapid innovation. Key trends include the rise of agentic AI, strong open source competitors from China, a burgeoning ecosystem of enterprise tool releases, and major shifts in how users organize and interact with AI-powered productivity suites.
NLW closes with a tease of a future deep dive into agentic tool infrastructure, promising continued coverage of these fast-evolving spaces.