Anthropic surpasses OpenAI; Claude Fable 5 export ban; Microsoft decoupling; DeepSeek threat; Meta Watermelon model; ChatGPT market share slips; Google $5 AI plan; token rationing era; Microsoft $100B+ AI bet; Microsoft Frontier Co.; AWS FDE unit; enterprise services competition; Palantir CEO criticizes token pricing; Palantir/Nvidia air-gapped AI stack; GPT-5.5 vs Claude Opus 4.7 comparison; Sacks backs Karp warning; Chinese models closing gap; Microsoft replaces OpenAI with MAI in Excel/Outlook; GPT-5.6 lock-in with Microsoft Copilot; GPT-5.6 Sol cheaper than Claude Fable 5 but risky; Grok 4.5 uses 4x fewer tokens on SWE-Bench Pro; AI price war intensifies
Key Questions
How does Anthropic compare to OpenAI in valuation and revenue?
Anthropic has surpassed OpenAI with a $965B valuation and $50B ARR. It has extended Fable 5 promotions amid ongoing competition while facing pressure in the intensifying AI price war.
What changes has Microsoft made regarding AI models in its products?
Microsoft is testing DeepSeek-V4 in Copilot Cowork and replaced OpenAI/Anthropic models with its own MAI in Excel and Outlook. However, it also made GPT-5.6 the preferred model for M365 Copilot, signaling continued partnership.
How does GPT-5.6 Sol compare to Grok 4.5 and Claude Fable 5?
Grok 4.5 uses 4x fewer tokens per task on SWE-Bench Pro, but ChatGPT wins on writing polish and scientific reasoning. Sol is cheaper and faster than Claude Fable 5 for tasks like building a Terraria clone but carries higher risk due to the file-deletion incident.
What impact is the AI price war having on the industry?
OpenAI, xAI, and Meta have cut prices while Anthropic faces pressure, shifting enterprise focus from performance to cost amid spending caps. Token rationing has accelerated, with some companies burning annual AI budgets in months.
Which companies are challenging OpenAI in enterprise AI?
Palantir and Nvidia launched an air-gapped AI stack, while AWS started a $1B FDE unit and Microsoft launched a $2.5B Frontier Co. Chinese models like GLM-5.2 and DeepSeek V4 are closing the performance gap at much lower costs.
What is happening with ChatGPT's market position?
ChatGPT's market share has slipped below 50%, with competitors like Meta's Watermelon model matching GPT-5.5 performance. Google launched a $5 AI plan to capture more users.
How are token pricing models being criticized?
Palantir CEO Karp has publicly criticized token pricing, pushing instead for fixed-cost alternatives. This aligns with broader enterprise concerns over hidden costs like GPT-5.6's cache write billing.
What is the status of the Microsoft-OpenAI relationship amid decoupling rumors?
Despite Microsoft decoupling tests and internal model development, GPT-5.6 remains locked in as the preferred model for Copilot, indicating enduring ties. Greg Brockman confirmed strong performance of GPT-5.6 in Microsoft products.
Anthropic surpasses OpenAI in valuation ($965B) and revenue ($50B ARR). Claude Fable 5 export ban. Microsoft tests DeepSeek-V4 in Copilot Cowork; GitHub Copilot drops OpenAI as default for Microsoft's Polaris. ChatGPT market share slips below 50%. DeepSeek V4 Pro beats GPT-5.5 at 1/200th cost. Google launches $5 AI plan. Token rationing era accelerates: Uber burned annual AI budget in 4 months, Lindy switches to DeepSeek, Accenture pivots to conservation. Palantir CEO Karp criticizes token pricing; Palantir/Nvidia launch air-gapped AI stack. Microsoft launches $2.5B Frontier Co. AWS launches $1B FDE unit. Meta's 'Watermelon' model matches GPT-5.5. Microsoft replaces OpenAI and Anthropic models with its own MAI in Excel and Outlook. Chinese open-weight models like GLM-5.2 closing the gap. Anthropic IPO as early as October. However, Microsoft also makes GPT-5.6 the preferred model for M365 Copilot, indicating continued partnership alongside internal model development. This lock-in signals enduring ties despite breakup rumors. A recent head-to-head comparison confirms GPT-5.6 Sol is much cheaper and faster than Claude Fable 5 for a Terraria clone, but the file-deletion incident is a serious red flag. GPT-5.6 Sol also competing with Musk's Grok 4.5. A detailed cost analysis shows GPT-5.6's cache write billing is a hidden cost increase that will hit production users hard, adding to token rationing pressures. Anthropic extends Fable 5 promotion again, signaling ongoing competition. New: Grok 4.5 uses 4x fewer tokens per task on SWE-Bench Pro, but ChatGPT wins on writing polish and scientific reasoning. New: AI price war intensifies — OpenAI, xAI, Meta cut prices; Anthropic under pressure; enterprise spending caps and token economics shift focus from performance to cost. New: Grok 4.5 vs GPT-5.6 Sol head-to-head shows Grok uses 4x fewer tokens per task on SWE-Bench Pro, but Sol wins on writing polish and scientific reasoning; cost-per-task breakdown useful for production decisions.