Descriptions:
Wes Roth covers three converging AI stories with significant market implications. Leading the episode is Anthropic’s confidential US IPO filing at a valuation just under $1 trillion — part of what could be an unprecedented year featuring IPOs from SpaceX, Anthropic, and OpenAI simultaneously. Roth frames the Anthropic filing as the AI industry’s first real balance sheet moment: revenue growth, inference costs, gross margins, enterprise customer concentration, and retention rates will all become public, giving both bulls and skeptics concrete numbers to work with for the first time.
On the model benchmarking front, Roth evaluates Claude Opus 4.8 against GPT 5.5 on the DeepSuite software engineering benchmark and ARC-AGI, noting that while Opus 4.8 has closed ground, GPT 5.5 currently leads on most coding metrics — with the caveat that Anthropic’s “ultra code” effort mode was not included in the third-party comparison. He also demonstrates a city-economy simulation he built entirely with Opus 4.8, featuring simulated workers, businesses, supply chains, and balance sheets, which he is developing into an independent LM benchmark targeting fluid intelligence rather than memorized problem-solving.
The episode closes with rumors that OpenAI is preparing GPT-5.6 in the near term, with some sources suggesting the coding capability leap is substantial enough to justify calling it GPT-6.
📺 Source: Wes Roth · Published June 02, 2026
🏷️ Format: News Analysis
