The creator clarifies that the positive experience he had with GPT-5 during early testing differs greatly from the downgraded, inconsistent version most users currently face due to OpenAI’s rollout strategy and auto router system, which has caused widespread frustration. He admits mistakes on his part, defends his integrity against accusations of bias, and calls for patience as improvements are made, emphasizing the model’s potential despite the flawed launch.
In this video, the creator addresses the widespread confusion and frustration surrounding the launch of GPT-5, clarifying that the experience most users are having now is significantly different and worse than the one he had during early testing. He emphasizes that his initial positive reaction was genuine and not influenced by any payment from OpenAI; in fact, he has incurred substantial costs using the model. He explains how he got involved with the GPT-5 launch video through personal contacts and chose to participate despite concerns about bias, aiming to provide an authentic perspective rather than marketing hype.
A major issue highlighted is the discrepancy between the model versions available to the public and the high-quality version he accessed via API during testing. The public primarily interacts with a downgraded or routed version of GPT-5 on platforms like chatgpt.com, which uses an “auto router” system to allocate requests to different model variants based on complexity. This system, intended to balance cost and performance, has led to a degraded user experience, especially for free and Plus users, causing many to feel misled by the initial hype. The creator also notes that the rollout involved disabling or hiding other model options, which further limited user choice and contributed to dissatisfaction.
He candidly admits to mistakes on his part, particularly releasing his launch video without monitoring community feedback due to being on vacation, which led to a mismatch between his experience and that of the wider audience. He also reflects on the timing of a critical video about Anthropic, which coincided poorly with the GPT-5 launch, fueling suspicions of bias or paid promotion. Despite the backlash and accusations of being a paid shill, he insists on his integrity, explaining that being honest has cost him financially and emotionally, but he remains committed to transparency.
The creator discusses technical challenges with GPT-5 integration, especially in tools like Cursor, where recent updates have seemingly degraded performance compared to his initial testing. He contrasts GPT-5 with other models like Anthropic’s Claude and Sonnet, noting that while GPT-5 can be superior in reasoning and cost-effectiveness, the current user experience is inconsistent and often disappointing. He stresses that GPT-5’s fundamentally different behavior requires new prompting and system design approaches, which were not adequately communicated or prepared for by OpenAI or third-party developers.
In conclusion, the video is a heartfelt attempt to clear up misunderstandings, defend the creator’s honesty, and call out OpenAI’s missteps in the GPT-5 launch. He acknowledges the model’s potential but criticizes the rollout strategy, especially the auto router and lack of clarity about model versions. He urges viewers to be patient as improvements are underway and hopes for a better experience soon. Ultimately, he expresses frustration at being misunderstood and punished for his authenticity but remains dedicated to providing honest insights into AI developments.