ChatGPT Pro 500

Original Article ↗ Hacker News Discussion ↗

OpenAI’s new ChatGPT Pro 500 plan, priced at $500/month, is being met with backlash as users note that usage allowances on the cheaper Pro tiers have been quietly reduced and that the company still refuses to publish clear, token-based limits. Many see the move as classic “shrinkflation”: paying more for less capability than the old $200 Pro 200 plan, with “ultrafast” access viewed as niche compensation rather than broad value. The changes are pushing some heavy users toward alternatives such as Anthropic, open‑weight models, or self‑hosted hardware, and raising broader concerns about opaque pricing, eroding trust, and the end of heavily subsidized AI access.

Pricing & Plan Changes

  • New tiers: Plus (~$20, 1x usage), Pro 100 (5x), Pro 200 (now 10x, previously 20x), Pro 500 (25x).
  • Many see Pro 500 as effectively a rebranded version of the old Pro 200 value, but at more than double the price.
  • Existing Pro 200 users are “grandfathered” at old usage until Oct 29, 2026, then drop to 10x at the same $200 price.
  • Some users report gradual, unannounced cuts to Pro 200 limits over recent months before this formal change.

Usage Limits, Ultrafast Mode, and Value

  • Core complaint: OpenAI no longer publishes clear token/message limits; marketing says only “more usage.”
  • Pro 500 includes an “Ultrafast” mode, ~8x speed but ~6x cost in internal quota, leading people to see it as an easy way to burn allowance.
  • Several say they hit weekly caps on Pro 200 doing intensive coding or autonomous research; Pro 500 still doesn’t feel “effectively unlimited.”
  • OpenAI is giving a one-time 62,500 “usage credit” grant (claimed $2,500 API value) to some subscribers; people are unsure how much real usage that is.

Trust, Transparency, and Regulatory Concerns

  • Many describe the changes as “shrinkflation” or “rug pull”: less usage for the same or higher price, after luring people into higher tiers.
  • Lack of explicit usage numbers and moving guardrails are seen as eroding trust.
  • Some argue this resembles predatory pricing: heavy VC-funded subsidies followed by sharp hikes once competitors are weakened; a few call for regulators/consumer-protection laws on opaque pricing.

Alternatives: Competitors and Open Models

  • Multiple commenters say higher prices push them to:
    • Competing closed models (Anthropic, Kimi, GLM, DeepSeek, etc.).
    • Open-weight models hosted by third parties or locally (GLM, Qwen, MiMo, etc.), especially as “flash” models get close to frontier quality for many tasks.
  • Others note subscriptions remain cheaper per token than API for heavy use, so they’ll “milk” them while they last.

Economics, Use Cases, and Sentiment

  • Some small businesses and high-earning independents see $500–$1,000/month as justifiable versus developer salaries, especially with 24/7 agents.
  • Others compare the cost to buying hardware (e.g., GPUs, Macs) or even hiring low-cost human labor and conclude it’s poor value.
  • Overall sentiment in the thread skews negative: people feel this marks the start of “enshittification” and the end of heavily subsidized AI access.