What’s changed (short and sweet)

OpenAI just gave ChatGPT a tune-up that’s trying very hard to be less dramatic and more useful. There are two flavors: GPT-5.6 Sol (the more focused model for paid tiers) and GPT-5.6 Luna (the one rolling out to free users). The goal? Quicker, clearer answers when you want them, and deeper, less noisy reasoning when you need it.

Highlights you’ll actually care about:

  • New control for how the model thinks: a reasoning slider you can nudge from “Instant” to “High”. Instant is speedy and short; High takes longer but digs deeper.
  • Sol has been trimmed to match casual chat use — fewer awkward detours, less unnecessary formatting, and smarter corrections instead of blind agreement.
  • Factual reliability got a boost: in OpenAI’s internal checks, Sol reduced responses with at least one factual error by about 68% compared with the previous version, and Luna showed a roughly 62% reduction in the same tests.

How this affects you (use it like a pro)

If you’re a paying customer, Sol is being pushed as the go-to for everyday and deeper tasks — think research, coding, complex planning — and it’s tuned to give the right level of detail without wasting your time. For quick stuff (“Who’s the president?”), expect single-line answers. For trickier questions (“How do US elections work?”), expect more context and a clear recommendation.

Free and low-cost users are getting good news too: Luna is becoming the default for those accounts and text-chat rate limits are being removed soon for those plans (other limits, like file uploads and image generation, still apply). Free users also get a new Think button to give Luna extra time on harder problems — tap it when the question needs more brainpower.

Two final bits to keep in mind:

  • The updated Sol is specifically for standard ChatGPT conversations — it doesn’t replace Sol in developer/agent products like Codex or Workflows.
  • OpenAI is rolling out extra safety boundaries for younger users, tightening rules around sexual content, dangerous activities, eating-disorder content, and graphic violence.

Everything’s being rolled out gradually, so if you don’t see the new models or controls right away, they should land in your account soon. Play with the slider, poke the Think button, and enjoy fewer tangents and more useful answers.