ChatGPT and Codex usage limits merging

OpenAI’s decision to merge usage limits for ChatGPT and its Codex/work tools is prompting concern that heavy coding or agentic workloads will now quickly exhaust quotas and block access to everyday chat. Many subscribers say the previous separation of limits was a key reason to stick with ChatGPT over rivals like Claude, Grok, and open‑weight models, and are now reconsidering providers or usage patterns. Commenters also criticize opaque, token-based metering and see the move as part of a broader trend of tightening limits and “enshittification” as AI platforms chase profitability.

Change announced

  • ChatGPT and Codex/Work will now share a single usage allowance instead of being separately metered.
  • “Instant” chat remains unlimited and does not count against usage.
  • Usage consumed will depend on model and task complexity; more advanced / “deep research” / agentic use burns more.
  • Users will see a combined usage meter and reset times; once exhausted, they must wait, upgrade, or buy credits.

Impact on workflows

  • Many had been using Chat’s effectively-unlimited tier for coding, repo work, and long reasoning specifically to avoid Codex quotas; this workaround is now closed.
  • Concern is high that running down Codex/agentic usage will also block basic Q&A and lightweight chat.
  • Several want a small reserved portion for plain chat or a user-controlled partition of the shared quota.
  • Some heavy users on Pro tiers note that if GPT‑6‑Pro / “thinking modes” now drain the same shared pool, this is a major downgrade.

Comparisons and switching behavior

  • Anthropic’s model (shared quota, 5‑hour + weekly limits) is cited as precedent; people disliked hitting limits there and appreciated ChatGPT’s previously separate/unlimited chat.
  • Some say this removes one of the main reasons to stick with OpenAI and are actively considering Claude, Grok, DeepSeek, Kagi, or local / open models.
  • Others argue top models from major labs are close enough that switching is easy when pricing or limits change.

Pricing, tokens, and fairness

  • Many see this as cost-cutting and “enshittification” after massive fundraising, with users paying more for less.
  • Frustration over opaque token-based quotas: users cannot easily know how many tokens their subscription buys, how caching/tooling affects burn, or when definitions change.
  • A few note that average chat uses tiny token volumes versus agentic coding, so metering chat for high-paying users feels unnecessary and counterproductive.

Abuse and design tradeoffs

  • Some argue merging limits is necessary because people aggressively exploited chat as a cheap coding/agent harness, undermining sustainability.
  • Others reject the idea this was “misuse,” saying they simply used the product as offered, accepting friction in exchange for lower cost.

Broader sentiment

  • Thread sentiment skews negative: fears of reduced utility, more friction, and eroding trust.
  • A minority see a “glass half-full”: this may spur competition and accelerate adoption of open or locally run models.