Reasoning effort: high. Has GPT-5 finally learned to think?
GPT-5’s new “reasoning_effort: high” model promises smarter, more reliable AI capable of handling complex tasks – potentially transforming industries and user experiences alike.
Image by Cybernews
GPT-5’s new “reasoning_effort: high” model promises smarter, more reliable AI capable of handling complex tasks – potentially transforming industries and user experiences alike.
If you had to give a sweet spot and a pain point for ChatGPT so far, what would you choose?
For me, I’d give kudos to its crate-digging abilities in recommending leftfield music or bringing some unorthodox recipes to the dinner table.
The thing that bugs me the most is definitely its tendency to out-talk – sputtering out weird flattery and excuses, especially when it’s accused of cutting corners.
But perhaps my grievances are about to change, especially after a GPT reasoning leak surfaced on July 19th.
More from Cybernews
Sainsbury’s installs more facial rec tech in stores: Should shoppers be worried?
"Phantom squatting” uses AI hallucinated domains for cyber attacks
What they’re doing to gamers is a robbery in broad daylight
Irish parliament expands Microsoft use, despite the EU’s efforts to ditch it
Can government AI actually scrub UAP footage from the internet?
The EU and Apple CEO Tim Cook held "constructive" talks after their Siri AI dispute in Europe
Deep thinking comes next
A lead engineer, Tibor Blaho, posted on X internal references to a model called “gpt-5-reasoning-alpha-2025-07-13,” with the timestamp suggesting a finalized build.
Tellingly, the code says "reasoning_effort: high" – strongly implying OpenAI is pushing for deeper, more structured cognitive output.
This is the most concrete signal yet that GPT-5’s public debut is imminent, possibly within weeks.
The leak aligns with past OpenAI behavior – model names often surface just ahead of official releases.
For example, gpt-4-turbo was first spotted in API logs before its Dev Day unveiling, and gpt-3.5-turbo-instruct appeared in playgrounds prior to formal rollout.
gpt-5-reasoning-alpha-2025-07-13
undefined Tibor Blaho (@btibor91) July 19, 2025
h/t @swishfever pic.twitter.com/Hq5csd0iC6
From blurting to planning
For the uninitiated, this points to a shift: GPT-5 may be less about speed or verbosity – and more about deliberation.
It could handle complex, layered tasks better – like “plan a weekend trip and fit it into my budget.”
Critically, it might reduce hallucinations by forcing the model to reason its way through prompts, instead of defaulting to confident guessing.
While ChatGPT isn’t as controversy-prone as Grok, I’ve still found it frustrating when it floods me with embellished or invented details.
Once it told me I’d planned holidays I never even thought of, I mean I’d love to bound my way through Mongolia, but I haven’t conceived the idea yet!
If your AI acted more like a chess player than a parrot, imagine the real-world potential – legal drafting, medical research, even therapy that doesn’t just serve up confirmation bias on demand.
A bit like some people really, sometimes it’s best just to zip it and get things done.
