Unitree founder predicts robotics breakthrough tied to GPT-6 or GPT-7

By Splitfeed EditorialLast updated Oct 13, 2026, 21:11 UTC

On October 12 and 13, 2026, the founder of Unitree, a Chinese manufacturer of quadruped and humanoid robots, publicly predicted that OpenAI's next major language model, variously cited as GPT-6 or GPT-7 depending on the outlet, could produce a transformative breakthrough in robotics within one to two years. The South China Morning Post and Tech Edition covered the remarks on October 12, pointing to GPT-6 as the catalyst, while TechRadar's October 13 piece cited GPT-7 and used the phrase "ChatGPT moment" to frame what the Unitree founder envisions: a sudden, large-scale leap in robotic capability analogous to the public impact of ChatGPT's launch. OpenAI has not confirmed a release timeline for either model.

The prediction matters because it comes from a hardware manufacturer with a direct commercial stake in the outcome, not a software researcher or investor. Unitree's robots are already deployed in research and industrial settings, which gives the founder's forecast more grounding than a purely theoretical claim. The core question the remarks raise is whether general-purpose language models are the missing ingredient for capable robotics, or whether the bottlenecks lie elsewhere, in physical hardware, sensor fusion, real-world data, and embodied training. The discrepancy between outlets on whether GPT-6 or GPT-7 is the cited model is a minor but telling detail: it suggests the prediction is tied more to a capability threshold than to a specific product.

No reactions from the Pro-AI, Anti-AI, or Middle Ground camps have been published yet on this story. Typically, Pro-AI voices would treat the prediction as consistent with a broader pattern of foundation models unlocking adjacent fields, pointing to the rapid capability gains seen between GPT-3 and GPT-4 as evidence the timeline is plausible. Anti-AI commentators would likely challenge the analogy to the original ChatGPT moment, arguing that language model gains do not automatically transfer to the physical constraints of robotics and that such predictions routinely overestimate the pace of real-world deployment. Middle Ground observers would probably acknowledge genuine progress in robot learning while questioning whether one or two years is a realistic window and stressing the gap between lab demonstrations and reliable commercial use.

The clearest near-term signals to watch are any announcements from OpenAI about GPT-6's release date or capability benchmarks, and whether Unitree or other robotics firms publish results showing large language model integration meaningfully improving robot performance outside controlled environments.