Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users

What changed

Source summary

OpenAI updated GPT-5.6 Sol for Plus/Pro users with more reliable facts and more focused answers (about 62-68% fewer factual errors), added a slider letting Plus/Pro users choose how much reasoning effort ChatGPT applies to a response, made GPT-5.6 Luna the default model for Free and Go users with a new "Think button" for deeper reasoning on complex questions, and enabled unlimited text chats for Free/Go users with added safety guardrails for users under 18 (covering romantic roleplay and age-restricted content).

Why it matters to your work

Practical interpretation

If you use ChatGPT Plus or Pro, expect noticeably more accurate answers from GPT‑5.6 Sol plus a new slider to dial reasoning effort up or down for a given task. If you use the free tier, you now get GPT‑5.6 Luna by default with unlimited text chats and a "Think button" for harder questions — worth another look even if you wrote off the free plan before.

When this could help

No action needed — the improvements apply automatically. Plus/Pro users should try the new reasoning slider. Free/Go users get Luna and unlimited chats automatically.

GPT-5.6 Sol now gives more direct answers scaled to the question, with about 68% fewer factual errors on financial/medical/legal prompts versus GPT-5.5 Instant. A new slider lets Plus/Pro users choose how much reasoning each answer gets. Free/Go users get GPT-5.6 Luna as default with unlimited text chats and a Think button for harder questions — Luna shows about 62% fewer factual errors than Instant.

How to get started

1. Plus/Pro: use the new slider to choose how much reasoning each answer gets. Free/Go: Luna becomes your default model automatically — use the Think button for harder questions that need more reasoning time.

   Provider guidance ↗

Build it into your routine

SCALA suggested workflow

1. Reasoning slider for efficiency

   For simple questions, keep the reasoning slider low for faster responses. For complex financial, medical, or legal questions where accuracy matters, turn it up.

What to check

• Sol's update applies only to the Chat experience — Work and Codex use the previous version. Think access for Free users is subject to abuse guardrails. Limits still apply to file uploads, images, and other tools. Accuracy figures are from internal evaluation, not independent testing. Source ↗