OpenAI’s massive language mannequin replace was lastly revealed to the general public on Thursday and shortly began rolling out to all customers. After the GPT-5 launch, OpenAI determined to discontinue its older fashions working ChatGPT like GPT-4o, GPT-4.1, o3, o3 Professional, o4-mini and extra.
For many use circumstances, GPT-5 takes over the reins from GPT-4o, the massive language mannequin which changed GPT-4 because the default mannequin on ChatGPT earlier this 12 months.
In case you’re questioning what precisely has modified in ChatGPT with the brand new mannequin, let’s take an in depth look.
GPT-5 vs GPT-4o: What’s new
1) Technical modifications:
OpenAI says that GPT-5 marks a giant leap in intelligence over all its earlier fashions, with the brand new mannequin stated to boast state-of-the-art efficiency throughout coding, math, writing, well being and visible notion.
OpenAI CEO Sam Altman even went on to tout GPT-5 as a “PhD-level skilled” throughout numerous domains.
2) Quicker responses and fewer hallucination:
OpenAI additionally claims that GPT-5 is far quicker and extra correct than its earlier fashions. In the meantime, the AI startup states that GPT-5 is much less more likely to hallucinate (make stuff up) in comparison with its earlier fashions.
GPT-5 responses have been 45% much less more likely to include a factual error than GPT-4o and 80% much less seemingly than o3 with net search enabled on anonymized prompts, OpenAI said.
It is value noting that whereas OpenAI has centered on pace and accuracy rather a lot with GPT-5, many customers on social media have claimed to obtain slower and inaccurate responses in comparison with GPT-4o.
3) Higher coding:
OpenAI says GPT-5 is its strongest mannequin for coding to this point, with enhancements in complicated front-end era and debugging bigger repositories. ChatGPT working GPT-5 can now create responsive web sites, apps and video games with simply easy prompts in pure language, lowering the necessity for customers to have any coding talents.
4) Higher writing:
GPT-5 can also be stated to be the ‘most succesful writing collaborator’ to this point by OpenAI, which the corporate says can translate tough concepts into compelling, resonant writing with literary depth and rhythm.
The social media critiques for GPT-5 for writing, nevertheless, appear to be diversified, with some customers reporting higher narrative consistency and improved rhythm and circulation, whereas others state that the brand new mannequin is much less inventive than GPT-4o and struggles with character consistency whereas giving shorter responses with much less emotional depth.
5) Well being associated questions:
OpenAI has made massive enhancements with GPT-5 in answering well being associated questions, which the corporate says will empower customers to be extra knowledgeable and advocate for his or her well being. The corporate states that GPT-5 acts extra like an lively thought associate by proactively flagging potential issues and asking questions to offer solutions to their health-related queries.
“The mannequin additionally now supplies extra exact and dependable responses, adapting to the person’s context, information degree, and geography, enabling it to offer safer and extra useful responses in a variety of eventualities,” OpenAI stated in its weblog submit.
6) Unified interface:
OpenAI has formally eliminated the mannequin picker from ChatGPT, that means that customers can now not select which mannequin they need to work together with. GPT-5 as an alternative comprises an effectivity mannequin and a reasoning mannequin, however switching between them is not as simple as earlier, with a real-time router now deciding which queries are despatched to the considering mannequin based mostly on the dialog sort, complexity, instrument wants, and question intent.
7) Utilization limits:
At launch, the GPT-5 utilization limits have been kind of unchanged for all customers, with free-tier customers being able to ship 10 messages each 5 hours, Plus customers having 80 messages each 3 hours, and Professional customers getting limitless messages for the usual model.
Nevertheless, with OpenAI going through quite a lot of flak for eradicating the older GPT fashions, the corporate has determined to quickly double the restrict for Plus customers to 160 messages each 3 hours.
As soon as the utilization restrict is breached, different queries of the person are then dealt with by a Mini model.
8) Decreased persona:
OpenAI has managed to scale back sycophancy (being too agreeable with customers) on ChatGPT with the GPT-5 replace, however long-time customers at the moment are complaining that the newest mannequin does not have a lot persona in comparison with the incumbent ones. Many customers who had grown accustomed to utilizing ChatGPT as their good friend, therapist, or only a companion of kinds state that the brand new mannequin lacks the emotional depth of GPT-4o and feels sterile, formal, and fewer participating.
Sam Altman has additionally commented on the matter and said that after the GPT-5 rollout to all customers, OpenAI would work on making the brand new mannequin “hotter.”
