the new GPT SOL 5.6 is unfair for plus users , increase the usage for plus by 2x - or a secondary ratelimits on small models

Open 💬 5 comments Opened Jul 30, 2026 by glook9001
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What issue are you seeing?

Shit ratelimits.

The new GPT-SOL 5.6 is unfair for Plus users; increase the usage for Plus by 2x—or a secondary rate limit on small models.

Most of the users now use GPT-SOL 5.6.

This leads to 30% less usage.

It's unfair that PRO users get 20x the usage of PLUS users.

Most of the data centers are charming to pro users.

Lower slightly the "GPU impact" those pro users are making—I demand 3x the rate limit. OR a daily limit—OR being limited to a free model?

Why do you crybaby on Twitter that LUNA is 80% cheaper now? Does it give me 80% more usage on LUNA? NO!

Rate limits are excessive—and I can't afford praying there is another reset—it's not economical; I can get afraid.

What steps can reproduce the bug?

I demand from OPENAI , that people will use less of the big models - or atleast a 5hours limit on SOL and TERRA only

if you need to remove the 5 hours limit --- remove the weekly limit, and make it a 2d or 1d limit

Its insanely annoying

What is the expected behavior?

_No response_

Additional information

_No response_

View original on GitHub ↗

5 Comments

glook9001 · 28 days ago

oh my fucking god, you give a reset, but the reset has lower token allowance in it
fucking great

github-actions[bot] contributor · 28 days ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #36053

Powered by Codex Action

alaestor · 28 days ago

This should be a discussion thread (if you insist...), not an issue.

Though, while I'm here, my two cents: 5.6 Terra is underrated, and so is Sol on low effort.

If you're hitting limits with casual usage then you you're probably just being careless with your model or effort picking, or abusing your context window.

If you're hitting limits with subagents or long-horizon looping/graph-like agentic tasks, or just vibecoding for hours a day every day, then you should probably be on the pro tier anyways.

glook9001 · 28 days ago
If you're hitting limits with casual usage then you you're probably just being careless with your model or effort picking, or abusing your context window.

the limits are insane

alaestor · 28 days ago
the limits are insane

I suggest you compare your token usage against API prices and maybe you'll feel differently. Codex sub limits are usually seen as quite generous. Unless there's something else going on (compromised session, a usage bug, etc) I think you're just being unreasonable.