Today's lead story
Same $20 plan: Claude gives you $1,178 of usage, ChatGPT gives you $211
First seen on X 33 hours ago@maria_rcks ♥ 5,194
SemiAnalysis published the results of buying and limit-testing subscription plans from Anthropic, OpenAI, Meta, SpaceXAI, MiniMax, Moonshot, Cursor, Cognition and others on October 5. The firm watched usage meters across thousands of calls, measured input, cached and output tokens separately, and priced the total at API rates.
A $20 Claude Pro plan bought roughly $1,178 of API-equivalent usage, against $211 for ChatGPT Plus. At $200, Claude delivered $11,726 and OpenAI $2,084. On mid-tier models (Opus 5.5 versus GPT-6.1 Sol) Anthropic's edge is more than 5x. A big factor is OpenAI's DevDay decision on September 29 to cut Pro 200 from 20x to 10x the Plus allowance starting October 30. On the most expensive top models, both labs were stingier.
Many called it not even close, and a post saying Opus 5.5 did irreversible damage to OpenAI spread. Others countered that Codex does more per token, so a straight comparison is hard. If you are deciding which subscription to keep, these are the numbers to look at first.