ChatGPT Pro

Original Article ↗ Hacker News Discussion ↗

Pricing and Perceived Value

  • Many see $200/month as a steep jump from $20 Plus, especially without clear, concrete “killer” workflows shown.
  • Others argue it’s reasonable for “tools for highly-paid knowledge workers,” comparing it to Bloomberg, CAD, legal research, etc.
  • Some individuals say they’d gladly pay $200 (or even more) because o1 already saves them hours on complex tasks; others say even $20 is hard to justify.
  • Several note this looks like classic price discrimination / anchoring: an expensive “Pro” tier makes $20 seem cheap and targets “whales” and enterprises.

Business Model, Costs, and Profitability

  • Commenters repeatedly mention reports that OpenAI is losing billions per year and needs to find sustainable revenue before 2029.
  • Some think $200 mostly reflects huge compute cost of o1-style “reasoning” (multi-H100 inference, long thought chains).
  • Others see it as a desperation or hype play amid intensifying competition and looming commoditization of models.

Capabilities and Real-World Use

  • o1/o1 Pro is praised for deep reasoning in math, debugging, system design, legal drafting, and analyzing complex configs or logs.
  • Several report that o1-preview outperforms junior engineers on some coding tasks, but still needs expert oversight.
  • Other users find the improvement over GPT‑4o or Claude 3.5 Sonnet modest, not 10×; some say Claude is still better overall.

Limits, “Unlimited” Use, and UX

  • Many are frustrated by opaque, dynamic limits in current plans and want clear quotas or pay‑per‑extra‑usage options.
  • Pro’s “unlimited” o1 access is viewed as a way to keep heavy users from creating many Plus accounts, but ToS forbids reselling, automation, or powering public services.
  • Some expect “soft throttling” or quality degradation under heavy usage.

Competition and Open Alternatives

  • Claude, Gemini, Perplexity, Amazon Nova, Qwen/QwQ, and local LLaMA-style models are frequently cited as cheaper or better for many tasks.
  • Several expect Anthropic and others to introduce similar high-priced tiers; others predict open models plus local compute will erode OpenAI’s pricing power.

Equity, Access, and Global Effects

  • Strong concern that $200/month effectively locks out most of the world (and many in the “rich West”), deepening the digital divide.
  • Some argue that if models stay closed and expensive, “digital serfdom” to a few AI corporations is likely; if open, poorer regions can run models locally.

Trust, Hallucinations, and Misuse

  • Persistent worry about hallucinations, overconfident wrong answers, and lack of clear confidence estimates.
  • Skeptics doubt LLMs “reason” in any meaningful sense and see Pro as “spicy autocomplete at a higher price.”
  • Others stress that the right mental model is “very capable but error-prone intern” and emphasize tests, reviews, and guardrails.