Subscription vs API pricing tested: Claude Max 20x gives double OpenAI's effective subsidy
PawelHuryn · x · 2026-10-02
Paweł Huryn deliberately burned portions of weekly allowances across popular plans and priced every model call at public API list rates, yielding the first concrete multipliers of "how much cheaper subscriptions are than APIs":
- OpenAI: allowance now proportional to price — Plus ($20/mo) is the baseline, Pro 100 ($100/mo) = 5x (tested), Pro 200 ($200/mo) = 10x (just downgraded from 20x).
- Claude: Max 20x ($200/mo) genuinely grants 20x baseline allowance, double the effective subsidy of OpenAI's Pro tiers.
- Meta Muse Code: High Usage ($15/mo) = 5x (not 3x), Power Usage ($50/mo) = 20x (not 10x), both tested.
All scripts and raw logs are open-sourced in the phuryn/experiments GitHub repo. This is part 1 of a 4-post thread.
More from Models
- Cloudflare open-sources decision model Clef, outperforming Jev within weeks — michellechen · 2026-10-02
- GPT-6.1 Sol nearly matches Astra at a quarter of the price on RareBench — danielmckinn0n · 2026-10-02
- New results from PostTrainBench v1.2 are in — mariofilhoml · 2026-10-02
- Opus 5.5 on /high burns through Claude's weekly limit in two days — rudrank · 2026-10-02
- VLM Run launches SystemOne API: typed, calibrated answers in one forward pass — spillai · 2026-10-02
- Document classification bake-off: VLM is ~7x more token-efficient than OCR — spillai · 2026-10-02