OpenAI is rolling out a extra dependable model of ChatGPT GPT-5.6 Sol for Plus and Professional customers, whereas Free customers are getting limitless textual content chats with GPT-5.6 Luna.
The adjustments are centered on making ChatGPT extra direct, factually correct, and constant throughout fast questions and deeper reasoning duties.
In our checks, BleepingComputer noticed that GPT now offers you better management over the mannequin’s reasoning and intelligence with a brand new slider.
Supply: BleepingComputer
As you’ll be able to see within the screenshot, you’ll be able to slide GPT’s reasoning from On the spot to Excessive. On the spot would reply nearly instantly, whereas Excessive reasoning might take a number of minutes.
You’ll be able to hold the slider low for on a regular basis questions or improve the reasoning effort for duties akin to analysis, coding, writing, planning, and sophisticated selections.
GPT-5.6 Sol is getting extra centered and dependable solutions
OpenAI says the up to date GPT-5.6 Sol has been tuned particularly for the way folks use ChatGPT, fairly than for the longer-running agentic workflows present in Codex and ChatGPT Work.
“It delivers extra centered solutions, adapts its stage of element to the query, avoids pointless formatting, and gives a useful correction when merely agreeing would not be helpful,” OpenAI stated.
For instance, if you happen to ask easy questions, GPT ought to reply instantly with out including pointless background. For extra difficult work, it ought to present sufficient element whereas holding the primary suggestion straightforward to determine.
A query like “Who’s the president of america?” could have a single-line reply, however a query like “How are elections held within the US?” could have an extended reply with background data.
OpenAI additionally claims that GPT-5.6 Sol is much less more likely to make factual errors when dealing with dates, numbers, sources, authorized guidelines, medical data, monetary questions, or assumptions.
In inside checks, OpenAI discovered that responses containing at the very least one factual error have been 68% much less widespread with GPT-5.6 Sol than with GPT-5.5 On the spot.
Furthermore, GPT-5.6 Luna decreased such responses by 62% in the identical analysis.
Free customers are getting limitless GPT-5.6 Luna chats
GPT-5.6 Luna will turn into the default mannequin for Free and Go customers this week, and a few customers are already seeing the change. I noticed it in my OpenAI account with a Go subscription that prices roughly $10.
Supply: OpenAI
Beginning subsequent week, OpenAI additionally plans to take away the speed restrict for textual content chats on these plans, though utilization will stay topic to abuse protections. Nevertheless, limits will nonetheless apply to file uploads, picture technology, and different ChatGPT instruments.
With a free account, you’ll nonetheless obtain a brand new Assume button that offers GPT-5.6 Luna further time to cause by means of more durable questions.
“For questions that require deeper reasoning, Free customers can faucet the brand new Assume button to provide GPT-5.6 Luna extra time to work by means of the reply,” OpenAI famous.
The replace doesn’t have an effect on GPT-5.6 Sol in Codex or Work
The revised model of GPT-5.6 Sol is designed for on a regular basis conversations and can solely be obtainable by means of ChatGPT’s common Chat expertise.
OpenAI can also be introducing further protections for customers believed to be below 18.
These embody stricter boundaries round romantic roleplay, sexual content material, harmful actions, consuming issues, body-image dangers, age-restricted items, and graphic violence.
At the moment’s adjustments are rolling out regularly.
Safety groups log 54% of profitable assaults and alert on simply 14%. The remainder transfer by means of your atmosphere unseen.
The Picus whitepaper exhibits how breach and assault simulation checks your SIEM and EDR guidelines so threats cease slipping by detection.
