SemiAnalysis: Claude plans deliver ~5x the API value of ChatGPT plans on mid-tier models

#What happened

SemiAnalysis's October 5 study (Andrew Megalaa, Max Kan, Dylan Patel) fed chunks of War and Peace into paid Claude and ChatGPT accounts, counting tokens between usage-meter movements until measurements narrowed to ±5%. On same-price plans it found Claude returns roughly five times the API-equivalent dollars of OpenAI plans when comparing the two labs' mid-tier models, Opus 5.5 and GPT-6.1 Sol — the daily-driver models both companies market hardest.

#Why it matters

OpenAI moved first. On September 29, Codex lead Thibault "Tibo" Sottiaux said the reopened $200 plan nets out at half the API-dollar value of the old one ("half the dollar in API spend compared to the old Pro $200 plan"); eligible existing Pro $200 subscribers keep their old limits through October 29. A separate price change cut GPT-6.1 Sol's cache-read price 50% while its $200-plan token limits stayed flat, lowering its API-equivalent value roughly 30%. Anthropic, by contrast, raised Opus allowances about 20% on Max and 50% on Pro with the 5.5 release.

The fivefold gap is a mid-tier result: on flagship models — Fable 5.1 and GPT-6 Astra — SemiAnalysis calls the limits "quite similar." And API-equivalent value is a dollar conversion of measured allowances, not proof every subscriber finishes five times as many tasks. SemiAnalysis also estimates subscriptions are only about 10% of Anthropic's revenue while consuming over 40% of its inference compute. OpenAI's counterargument, per Sottiaux: cheaper, more efficient models should let developers get more done than a month ago, and OpenAI Pro plans carry no five-hour cap — though SemiAnalysis says the missing cap doesn't offset the Opus advantage.

Verified October 6, 2026.

#Sources

• Implicator.ai
• AI Weekly