GPT-5.6 Sol gets a ChatGPT-specific tuning pass, not a new Work or Codex model

OpenAI says it has updated GPT-5.6 Sol for Plus and Pro users in the ChatGPT conversation product. The stated changes are more focused answers, greater factual reliability, and a slider that lets a user choose how much thought ChatGPT puts into a response. This is a product-surface update to an existing model experience, not evidence of a new model release or a change to the API offering.

The important architectural boundary is the one OpenAI draws itself: the revised Sol is optimized for everyday chats and is available only in ChatGPT's Chat experience. OpenAI also says the same model now underlies both Instant responses and deeper reasoning for Plus and Pro users, with the effort slider available on web, mobile, and desktop. The company states that its Work and Codex versions of GPT-5.6 Sol are not changing in this update.

OpenAI compares the new ChatGPT behavior with GPT-5.5 Instant using an internal evaluation of prompts in financial, medical, and legal domains that required factual detail. It reports that responses containing at least one factual error were 68% less common for GPT-5.6 Sol than for GPT-5.5 Instant. That is a vendor-reported result from a bounded internal evaluation, not an independent benchmark or a guarantee about accuracy in every workflow; users should treat it as evidence about the tested setting rather than a general safety claim.

The companion access change concerns GPT-5.6 Luna, which OpenAI says will become the default for Free and Go users, with unlimited text chats and a Think button scheduled to follow subject to abuse guardrails. File uploads, images, and other tools retain limits. For technical users, the practical conclusion is narrow: evaluate the Sol update in the ChatGPT interface if response focus or adjustable effort matters, but do not infer a behavioral change for Codex, Work, or developer integrations from this announcement.