OpenAI may be preparing to revisit one of the most important questions in AI development: how much of its work should be open source. In a recent Reddit Q&A, Sam Altman said the company may have misread the direction of the field and should work out a new approach.
The comments also came with a more restrained view of OpenAI's competitive position. Altman said the company expects to keep improving its models, but not with the same distance from rivals that it previously enjoyed.
A possible turn on open-source AI
Altman's clearest signal concerned OpenAI's open-source strategy. He acknowledged that OpenAI is "on the wrong side of history" on the issue and said the company needs to figure out a different path.
That does not mean a new policy is imminent. According to the source article, internal conversations are taking place, but Altman also said not everyone at OpenAI shares his view. He added that the matter is not the company's top priority right now.
Even with those limits, the statement matters because it points to a tension at the center of OpenAI's current position. The company grew from a background tied to open-source development, yet its most important model work in recent years has largely remained closed. Altman's comment suggests that OpenAI's leadership is at least reexamining whether that posture still fits the market and the technology landscape.
The lead is expected to shrink
Altman also framed OpenAI's future with less dominance than in earlier periods. He did not say the company would stop advancing its models. Instead, he described a market where OpenAI improves, but competitors close some of the gap.
"We will produce better models, but we will maintain less of a lead than we did in previous years," Altman said.
That is a significant expectation for users, developers, and businesses watching the AI model market. It suggests that OpenAI still sees itself moving forward, but no longer assumes the same degree of separation from other teams. In practical terms, a narrower lead could make model choice less obvious and increase pressure around pricing, features, transparency, and deployment options.
The source article does not say which competitors Altman had in mind. It does, however, connect the broader discussion to Deepseek and the performance of open-source models. The implication is that closed models and open-source models are now part of a much closer contest than many observers may have assumed.
Product details remain limited
Altman did not give a timeline for GPT-5. That leaves one of the most closely watched next steps in OpenAI's model roadmap unresolved.
OpenAI does plan to keep updating its GPT-4o series. The source article also notes that OpenAI product manager Kevin Weil said the native GPT-4o image generation capability shown last May is still several months away, while remaining in development.
The full o3-model has a more defined, though still broad, window. Altman described it as "more than a few weeks, less than a few months" away. That phrasing leaves room for uncertainty, but it confirms that the model is still expected within that general period rather than being positioned as a distant project.
Together, these details show a company still building across several tracks: GPT-4o updates, image generation work, and the o3-model. What is missing is a firm schedule for GPT-5 or a detailed public plan for any open-source shift.
Transparency becomes part of the competition
OpenAI also plans to make the thought processes of its o-models more transparent, following Deepseek's example. That is another sign that competitive pressure is not only about raw model performance. It is also about how much users can see and understand when a model works through a task.
Kevin Weil said OpenAI must balance user interests with the risk that competitors could reverse engineer its technology. That tradeoff sits close to the open-source debate. More transparency can help users trust and inspect model behavior, but the company also wants to protect technical advantages it considers important.
The source article does not detail how this added transparency will work. It only says that OpenAI intends to move in that direction while managing the competitive risks. That makes the plan notable, but still incomplete from a user's perspective.
OpenAI's roots and the current debate
The question of open source has followed OpenAI for years. The source article says the company's relationship with open source changed substantially after GPT-3 and even more after GPT-4. Since then, OpenAI has mostly stepped away from open-source development, while still releasing smaller projects such as the Whisper speech-to-text model.
Altman previously pointed to safety concerns as the main reason for that shift. The source article says he argued that the gap between commercial and open-source development was growing because of safety considerations.
At the same time, Altman had already said back in May 2023 that closed and open-source approaches could be in a close race. That makes the current reconsideration less like a sudden reversal and more like a debate that has been building for some time.
For now, OpenAI has not announced a concrete new open-source policy. The most important takeaway is narrower but still meaningful: Altman is publicly questioning the company's current stance, expecting a smaller competitive lead, and pointing toward more transparency for o-models while product timelines remain selective.