Particle.news
Download on the App Store

OpenAI Removes ChatGPT Text Limits and Rolls Out Tuned GPT-5.6 Variants

OpenAI cites internal tests showing far fewer factual errors as it lifts text-chat caps to broaden access.

Overview

  • OpenAI made a ChatGPT-tuned GPT-5.6 Sol available to Plus and Pro subscribers immediately on Thursday, August 6, 2026, and added a reasoning slider that lets users trade speed for deeper, slower answers.
  • Free and Go users will get GPT-5.6 Luna as the default model and staged access to unlimited text chats plus a new Think button for harder questions this week and next week.
  • In internal evaluations reported by the company, factual errors were 68% less common with the tuned Sol and 62% less common with Luna compared with GPT-5.5-Instant.
  • OpenAI said it will keep separate usage limits for files, images, voice, and image generation, and that the Sol instance used by ChatGPT Work and Codex will not change as part of this update.
  • The change follows ChatGPT’s recent growth to roughly 1 billion weekly users and could shift how millions use conversational AI by lowering the barrier to extended text interactions and giving builders finer control over response effort.