Alexander Lewis
Mar 21, 2024
OpenAI's eagerly anticipated GPT-5, the next iteration of its Large Language Model (LLM), is reportedly nearing completion, with insiders suggesting a launch within the coming months, according to sources close to Business Insider.
Unnamed informants disclosed to the publication that GPT-5's release may coincide with the summer season, with OpenAI currently circulating demonstrations of the technology and updates to its associated ChatGPT chatbot to enterprise clients for evaluation.
Among those who witnessed a GPT-5 demo tailored to their company's needs, one CEO expressed high praise for OpenAI's advancements, describing the experience as "materially better" in conversation with Business Insider.
Read more: Google Gemini — Everything you need to know
Internal presentations by the brand have also hinted at unrevealed features of GPT-5, notably an AI agent capable of autonomous task execution without human intervention.
While enterprise partners undergo internal testing of GPT-5, sources indicate that OpenAI continues its training for the forthcoming LLM. The release date hinges on this ongoing process, including safety assessments such as red teaming — a cybersecurity practice where simulated attacks are launched to uncover potential vulnerabilities before public release.
Amidst this process, OpenAI has refrained from setting a definitive launch date for GPT-5, acknowledging that predictions could shift as safety testing progresses.
The last official update from OpenAI regarding GPT-5 dates back to April 2023, when the company stated no immediate plans for training. CEO Sam Altman dismissed early speculation about GPT-5 shortly after the release of GPT-4, labelling such rumours as "silly." However, reports since then have hinted at the completion of training in 2023 and a potential launch in 2024, sparking anticipation within the tech community.
GPT-5's deployment is anticipated to cater primarily to OpenAI's enterprise clientele, who constitute a significant portion of the company's revenue.
There are speculations that the launch could introduce tiered models akin to Google Gemini LLM tiers, tailored to various purposes and clientele. Presently, GPT-4 and GPT-4 Turbo serve as the backbone for ChatGPT Plus, a paid consumer tier product, while the earlier GPT-3.5 model powers the original ChatGPT chatbot, which remains free to use.
Over the past year, users have reported instances of GPT-4's perceived "laziness" or "dumbing down," attributing it to various factors such as ongoing training efforts, resource limitations, or potential parallel training of newer LLMs alongside existing ones.
Rumours about hypothetical LLMs, tentatively labelled GPT-4.5 and GPT-4.5 Turbo, circulated in December 2023, accompanied by leaked information regarding pricing tiers and advanced multimodal capabilities. Altman, when queried about the leaks, dismissed them casually on social media.
