To cap a day release of the product, researchers of Openiai, engineers, and executives, including the Openi CEO of COO in a The width ranging Reddit Father on Friday.

OpenI company finds itself in a small position. It struggles with understanding that it is a Ceding Ground in the AI ​​race Chinese companies like dereseeekwhere opukii says maybe the IP is probably stolen. Chatgppt makes the Show relationship with Washington and simultaneously pursuing a Ambitious Data Center projectWhile reported Set up Groundwork for one of the largest financing phase of history.

Altman admits that Dereseeh deceives Openii’s lead in AI, and he also said that he believes that the opening “in the wrong side of history” to open its technology. While Openi has models of open models before, the company is generally preferred by a proprietary, closed mode of source development.

“(I personally think we need) to know the different open source strategies,” said Alman. “Not all of Openi share this view, and it also doesn’t have the current highest priority (…) we will make better models (keep a lead than before.”

In a follow-up response, Kevin Weil, the principal product officer in Openii, says the OpenI considering older older models that will never end. “We really think about doing more about it,” he said, with no more details.

More prompting the openi to think about its release, Almeny says the company has revealed about how to release rational models, like this modeling model now, show their “process of thinking. ” Today, openii models hide their reasoning, a strategy intended to prevent competitors from scraping training data for their own models. On the contrary, the model of reasoning in Dereeseek, R1, showed the entire chain of mind.

“We’re working on showing a bunch more than we’ve shown today – (shows the model’s thinking process) is so easy,” In addition. “TBD all – showing all the chains of thinking lead to competitions, but we also know people (at least users of power) so, so we can find the right way to balance it. “

Altman and Weil tried to attack gossip Chatgpt, the chatbot app by which Openi launched most of its models, increasing its price. Altman says he wants to make chatgpt “cheaper” over time, if possible.

Altman used to say the money of Openi with the best chatgt forces, chatgpt pro, worth $ 200 per month.

In a relatively relevant thread, it says Openi continues to see evidence that greater power leads to “better” and more models. Thats the big part what projects are like stargate, openii’s recently announced Great projects at the Center Center, Weil said. Serving a growing user base is to fuel computation within Opui, as well, he continues.

Asked about the development of one’s own progress that can be used in powerful models, Altanman said he thinks a “rapidly withdrawal” is more than ever. Self-development progress is a process by which a AI system enhances its own intelligence and capabilities with no human input.

Of course, it is worth noting that Almman is famous for overwhelming. She is not long ago lowed the bar on for the through.

A reddit user asked if OpenII models, self-reliance or not, use to develop weapon weapons – specified nuclear weapons. This week, Opuii announced an association with the US government to give models of National Nuclear Nuclear Laboratories.

Weil said he trusted in the US government.

“I know these scientists and their experts in AI are in addition to world class researchers,” he said. “They understand the power and limits of models, and I don’t think that any chance they are just a model output to a nuclear calculation. They are evidence-based and evidence of their experimenting and data work to validate all their jobs. “

The Opeai team was asked many questions to a more technical character, which if the next model of openii, had to be released (“more than a few weeks,” said Alman, if the next few months, the model of non-reason, GPT-5, may have land (“no one timeline,” as the altman), and if Openi can open one heirs at Dall-e 3, released about two years ago, the highly old man’s tech. Henree’s clock is improved by jumps and bounds since Dall-E 3, and the model no longer competition in many benchmark tests.

“Yes! We’re working on it,” Weil says a Dall-E 3 follow-up. “And I think it can wait.”