ChatGPT Pro now starts at $100/month

Original Article ↗ Hacker News Discussion ↗

OpenAI has introduced a $100/month ChatGPT Pro tier that sits between the existing $20 Plus and $200 Pro plans, offering higher usage limits and access to the GPT‑5.4 Pro model. Commenters debate whether this pricing narrows the gap with rivals like Anthropic’s Claude, with many claiming GPT‑5.4 and Codex are now superior for complex engineering work while others still prefer Claude for planning and creativity. The change is also read as a sign that the era of heavily subsidized AI access is fading, amplifying concerns about rising costs, opaque rate limits, data privacy, and growing dependence on a few dominant providers.

New Pro Pricing & Plan Structure

  • ChatGPT now has three tiers: Plus at $20, Pro at $100 (“5x usage”), and Pro at $200 (“20x usage”).
  • The $100 plan is new; $20 and $200 already existed. Pro tiers include access to GPT‑5.4 Pro; Plus does not.
  • A temporary promotion doubles Pro usage (effectively 10x/40x vs Plus) until a stated date.
  • Several people find the “From $100” wording and “5x or 20x more usage” copy confusing or borderline deceptive, especially since the $200 option is hidden behind a click‑through.

Usage Limits, Subsidies, and “Enshittification” Concerns

  • Multiple reports claim Plus usage limits feel tighter now (some mention ~⅓ of earlier), possibly so Pro can be marketed as 5x/20x Plus.
  • Others note a past 2x Plus promotion ended, so limits only feel lower.
  • Debate over whether this marks the end of heavy subsidization and the start of moving users up‑market toward $100+ plans and ads on lower tiers.

Model Quality Comparisons

  • Strong split opinions on GPT‑5.4/Codex vs Anthropic’s Claude Opus and Google’s Gemini Deep Think:
    • Many argue GPT‑5.4 (especially xhigh) is significantly better for systems programming, debugging, reverse‑engineering, math, and deep reasoning, albeit slower.
    • Others find Opus better for planning, creativity, ideation, and “friendlier” interaction, but more error‑prone on hard technical tasks.
  • Several people say Codex/GPT catches numerous bugs in Claude‑generated code; others feel both ecosystems are roughly on par, just with different “styles.”

Developer Workflows & Tooling

  • Heavy discussion of using Codex vs Claude Code vs GitHub Copilot.
  • Some prefer Copilot as the most cost‑effective, IDE‑integrated option; others lean on Codex for deep context with local tools.
  • Concerns about “agentic development” producing unreviewable volumes of code and technical debt; tension between speed gains and human review capacity.

Ethical, Trust, and Privacy Debates

  • Several refuse to pay OpenAI due to distrust of leadership, perceived dishonesty, monopoly ambitions, and data‑use concerns.
  • Counterpoint: many users prioritize capability over ethics, arguing all major AI vendors are similarly profit‑driven.

Market Dynamics and Astroturfing Suspicions

  • Some see the $100 tier as a response to competition (especially Claude) or slowing growth.
  • Others note explicit statements that it was added for “medium‑usage” demand.
  • A subset suspects active astroturfing in pro‑AI and pro‑vendor comments, given the stakes.