TBPN
TBPN

Meta’s AI Comeback Moment, Claude Mythos | Diet TBPN

April 9, 2026

AI Summary

5 min read

Meta released a new AI model called Muse Spark, its first major model in over a year, and the stock jumped nearly 8% on the news. The model is a closed, not open-source, system—a strategic shift that was predicted years ago and marks a potential turning point for the company's AI ambitions. Meanwhile, Anthropic unveiled Mythos, a model so capable at finding zero-day exploits that it is being gated to only fifty critical infrastructure companies, sparking both excitement and skepticism about whether the safety rationale is genuine or a marketing play.

Meta’s Closed-Source Pivot and the Muse Spark Launch

Meta’s new model, Muse Spark, is its first major AI release since Llama 4 in April 2025 and its first that is not open-weight. The model scored 886.4 on the Artificial Intelligence Analysis Index, behind only Gemini 3.1 Pro, GPT-5.4, and Claude Opus 4.6. On internal benchmarks, it outscored Google Gemini on some tests and was competitive with OpenAI and Anthropic models on others, while significantly outperforming xAI’s Grok on most tests. The stock rose nearly 8% on the announcement, reflecting relief that Meta’s massive AI spending is producing a credible product.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

What you'll learn

  • 1 (00:00) **Meta Launches Muse Spark, a New Closed-Source AI Model** - Meta's first major AI model in over a year, Muse Spark, is a closed-source model, marking a departure from its open-source Llama strategy.
  • 2 (03:18) **Muse Spark Details and Benchmark Performance** - The model, likely codenamed "Avocado," shows mixed benchmark results, outperforming on some (e.g., Health Bench Hard) but underperforming on others (e.g., ARC AGI-2).
  • 3 (07:45) **Meta's AI Strategy: Internal Use vs. External Sales** - The hosts discuss Meta's economic incentive to build its own model to reduce reliance on expensive third-party APIs like Anthropic's Claude.
  • 4 (10:24) **Market Reaction and Future Model Plans** - The stock is up ~8% as Muse Spark is seen as a "close to frontier" model, easing investor uncertainty about Meta's massive AI CapEx.
  • 5 (13:21) **Anthropic's Mythos Model: A Cybersecurity Breakthrough** - Anthropic's new model, Mythos, is exceptionally good at finding zero-day exploits and is initially only available to ~50 critical infrastructure companies.
  • 6 (15:37) **The "Software-Only Singularity" and Skepticism** - The hosts contrast Mythos's immediate utility in cybersecurity with slower-progress fields like biology, which require real-world feedback loops.
  • 7 (18:56) **The Next Model Cycle Begins: X.AI and More** - Elon Musk announces X.AI is training seven new models, including a 10-trillion parameter model, signaling the next aggressive cycle of model releases.

+ Full timestamped outline available in the app

Guests on this episode

Show Notes

Diet TBPN delivers the best of today’s TBPN episode in 30 minutes. TBPN is a live tech talk show hosted by John Coogan and Jordi Hays, streaming weekdays 11–2 PT on X and YouTube, with each episode posted to podcast platforms right after.


Described by The New York Times as “Silicon Valley’s newest obsession,” the show has recently featured Mark Zuckerberg, Sam Altman, Mark Cuban, and Satya Nadella.


Follow TBPN: 

https://TBPN.com

https://x.com/tbpn

https://open.spotify.com/show/2L6WMqY3GUPCGBD0dX6p00?si=674252d53acf4231

https://podcasts.apple.com/us/podcast/technology-brothers/id1772360235

https://www.youtube.com/@TBPNLive

TBPN

More from this podcast

TBPN →