TMTPOST -- OpenAI seems ready for the next-generation large artificial intelligence (AI) model while escalating the price war.
Related articles
Screenshot from OpenAI Podcast
OpenAI CEO Sam Altman implied the company’s next flagship large language model (LLM) would hit the market this summer. Asked about the timeframe of GPT-5 during the startup’s newly-launched official podcast on Wednesday, Altman said the model is coming “probably sometime this summer”, adding that the exact timeframe is not clear yet.
GPT-5 is expected to integrates many of OpenAI’s advanced AI capabilities, particularly in areas like multimodal processing, improved reasoning, advanced voice model, media generation and enhanced memory. Altman in February said OpenAI’s two primary lines of models--the traditional LLMs, or GPT models, and reasoning O models would combine into a single line of models, starting with GPT-5.
Altman on Wednesday indicated GPT-5 may represent more than a performance upgrade. It could be deemed as OpenAI’s first real stride toward unified, agent-like models, moving closer to its artificial general intelligence (AGI) vision. “I think we are near the end of this current mountain,” he said. “I am excited to just get to GPT-5, GPT-6, and I think that’ll be easier for people to use.”
OpenAI is internally debating whether to go with GPT-5, then 5.1, 5.2, or keep doing updates like GPT-4o to avoid confusion. Altman suggested the company could alter the naming conventions. “Should we just keep calling those GPT-5, right? GPT-4o? Or should we call those 51253?” he asked.
As a single, integrated system with advanced features like multimodality, memory and reasoning, GPT-5 allows users not to toggle between specialized models. Altman stressed its ability to personalize interactions and boost responsiveness over time. “More sophisticated memory is probably my favorite recent feature,” he said.
Altman hinted he is open to ads on ChatGPT, stating he's "not totally against" the idea-a shift that could reshape how the chatbot is monetized. But he warned it would take "a lot of care" to get the experience right. He emphasized that modifying the model's output based on who pays for the ad would be "a trust-destroying moment" for its users.
Wednesday saw OpenAI was reported to offer its major business customers deep discounts. OpenAI has started selling ChatGPT at discounted rates when customers agree spend money on additional AI products, The Information cited a spokesperson for the startup and executives at large companies.
OpenAI reportedly discounted ChatGPT Enterprise 10% to 20% for customers which sign a multiyear agreement to buy other AI tools, including a bespoke version of its Deep Research agent and its Codex coding agent. Customers can also get discounts if they commit to spending a certain amount on OpenAI’s application programming interface (API) to its models, according to the report.
The reported discounts were part of OpenAI‘s efforts to expand its customer base and challenge rivals, including its largest shareholder Microsoft Corporation, as it seeks dominance in enterprise AI adoption. Microsoft salsespeople have lost some deals because they didn’t get approval to match OpenAI’s discounts, per the report.
OpenAI earlier this year told investors ChatGPT Enterprise had hauled $100 million in revenue last year, and it anticipated the Enterprise product would generate $400 million for this year with 2 million subscribers. Revenue for the next year is projected to be $1 billion, and the annual revenue would come in at $6.2 billion for 2030, with 16 million subscribers, up from 4 million in 2026.
更多精彩内容，关注钛媒体微信号 （ID：taimeiti），或者下载钛媒体 App
