Anthropic/OpenAI's Best Out, Also 'Best of Rest' from Meta & SpaceX. AI-RTZ #1144

AI: Weekly Summary. RTZ #1144

…week ending Friday July 10, 2026

1. Anthropic & OpenAI top AI models ‘Freed’, but far from free: US Government green-lit OpenAI latest GPT 5.6 model globally, to now compete with Anthropic’s Fable 5 top models. Both top tier US frontier AI models are free to access around the world. But the pricing is far from free, with both moving aggressively to a la carte pricing, outside their top tier ‘all you can eat’ subscription tiers. This is of course to ramp revenues as fast as possible, before both company’s anticipated mega-AI IPOs later this year and next. The green light followed Commerce’s Center for AI Standards and Innovation testing, with OpenAI experts stationed in D.C.; GPT-5.6’s flagship ‘Sol’ (plus Terra and Luna tiers) went wide Thursday, joining Fable 5, redeployed July 1 — while next-gen models like Mythos stay gated case-by-case. It’s the ‘stop and go’ policy I covered on ARD #109, as China’s open-source models race around the logjam. More here.

   

2. Meta & SpaceXAI’s latest out as well: Both Meta and SpaceXAI now have their latest, ‘best of the rest’ AI models out as well this week. Meta’s Muse Spark 1.1 — the first real fruit of its multi-billion-dollar Superintelligence Labs effort under Alexandr Wang — is a ~1.5-trillion-parameter model, stronger on agents, and notably its first serious pay-to-use API at ~25% of the top models’ cost — the monetization arm of the ‘ZWS’ cloud ambitions I’ve flagged. SpaceXAI’s Grok 4.5 draws on its $60 billion Cursor deal (yet to close in Q3), stronger on coding, legal and finance. Both are the US ‘best of the rest’ at ~1.5T — running for second place behind Fable 5 (rumored 5-10T, “the first of the next generation”) and GPT-5.6 (top of the current one), competing on cheaper, more efficient pricing. How many ‘good enough’ models can the market price, absorb, digest, and use at once? For now, competition in the market is a good thing. More here.

   

3. Post-mortem on Anthropic ‘Blip 2.0’ & Implications: Three weeks later, we now have more detail on how the US Government driven ‘the Blip 2.0‘ went down at both Anthropic and OpenAI. And the lessons are worth understanding for implications going forward. Axios’s behind-the-scenes traces the 20-day showdown back to an Amazon ‘jailbreak’ alarm — later disputed by cybersecurity experts in an open letter — a June 12 Howard Lutnick call to Dario Amodei, engineers deployed to D.C., and gradual multi-agency sign-off to a July 1 release. Treasury’s Scott Bessent was first to hear of it; with ‘AI Czar’ David Sacks departed, Lutnick now steers the US AI ship. The unresolved part: there’s still no clear framework, transparency or timelines for approving future models — and the latest frontier models remain out of reach for most. Brad Smith, Microsoft’s veteran DC emissary, had some timely thoughts on how the US Government could do things differently going forward. More here.

   

4. Alibaba case study on open to closed AI models: China’s leading tech and AI company, Alibaba, is a timely global case study on managing the success and economics between top open source models, and monetizing them in a more closed manner. A transition that is also underway in the US with Meta with its Meta Superintelligence Labs investments. Alibaba’s Qwen became the world’s most-downloaded open-source model — ~1 million downloads a day on Hugging Face — yet AI drove just $1.3 billion (under 4% of revenue) last quarter against a $55 billion+ infra plan. The NYT reports the Qwen team fractured over commercialization (lead engineer Lin Junyang and others departed), and Alibaba is now steering customers toward closed, paid models — the same Llama-to-Muse-Spark pivot playing out in the US. As one investor put it, “there isn’t an AI company with a sustainable business model right now.” Alibaba’s stock is down 37% this year. A global ‘canary in the coal mine’ on this open/closed issue indeed. More here.

   

5. New China metrics on ‘RAMageddon’: As expected, the ‘RAMageddon‘ impact of higher memory costs and supplies are being seen in the latest smartphone sales figures out of China. Counterpoint Research data shows China smartphone sales fell 13% during the month-long 618 festival as brands raised prices to offset memory costs — Honor down 33%, Xiaomi down 24%, with Huawei the only major brand to grow (+19%). Apple slipped 9% but climbed to #2 on iPhone 17 Pro discounts (up to $295 off), its supply-chain moat intact. Counterpoint now projects a double-digit China shipment decline for the year — and the same drag is coming for PCs, laptops and all gadgets, straining the AI ‘mainframe-to-local’ barbell I’ve long tracked. More here.

   

Other AI Readings for weekend:

1. Anthropic’s fascinating AI Research into ‘J-Space’, on how AI models mathematically emulate human neuron driven thinking. More here.

2. The latest ‘Sum of the Part’ (SOTP) analysis on Microsoft, as industry spin-offs are announced. More here.

(Additional Note: AI Ramblings Daily on YouTube, is now a weekday Daily podcast called AI Ramblings Daily (ARD). Just passed the #100 podcast milestone mark. Different content than AI-Reset to Zero (AI-RTZ) substack, which remains a daily morning substack write-up. With now over 1140+ ‘MY TAKES’ on key AI events and issues turbulently flowing by. ARD is typically a 20 minute afternoon podcast every day, on my take on additional AI developments. Both daily substack and podcasts typically discuss different AI issues and items. There is a daily text summary of the daily podcasts here on the substack, as well as one minute YouTube ‘Shorts’ video clips on the key topics discussed. And all are free to subscribe for now. Try this week’s series with ARD Episodes 112 — AI Chips Not Easy, 113 — What’s in a Name, 114 — AI Price Umbrellas, 115 — US-China Cross-Purposes, and 116 — Best of the Rest here):

Up next, the Sunday ‘The Bigger Picture’ tomorrow. Stay tuned.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)