Summary
Theo digs into OpenAI’s latest change to Codex subscription limits and what it means for developers choosing between OpenAI and Anthropic. After a rough stretch for Codex fans, including reduced limits on the $200 plan, OpenAI announced that usage limits are doubling across Pro tiers at $100, $200, and $500, as well as business premium.
Theo traces the backstory, including OpenAI’s September announcement that the $200 tier would net out to roughly half the API-dollar value of the old plan, and the reasoning that cheaper API pricing meant subscription value had to be rebalanced. He also flags that bundling changes may give some users less usage than before.
The core of the video is a value comparison. Rather than looking only at per-token prices, Theo breaks effective pricing into three factors: token prices including caching, how many tokens a model needs to solve a task, and what each subscription actually delivers. Using Artificial Analysis data, he compares GPT-6.1 Soul and Claude Opus 5.5 on intelligence scores and cost per task, and cites Adam Holter’s subscription-adjusted numbers showing the gap narrowing at the $200 level.
He wraps up with guidance on which plan makes the most sense, making it a useful read for anyone weighing Codex against Claude Code.
📺 Source: Theo – t3․gg · Published November 03, 2026
🏷️ Format: News Analysis
