GPT-4.5

GPT-4.5 is a short but telling storyline on ToTheMoon: a case of OpenAI shipping a model whose place in the lineup neither the company nor its users could explain. The hosts treat it less as a technology than as a product with a price and a purpose, and ask who needs a model that costs orders of magnitude more than the default when its advantage cannot be shown on ordinary tasks.

The main analysis came in March 2025. In episode 048 the hosts noted that GPT-4.5 had been expected to combine the strengths of 4o and the reasoning models, and instead arrived with no obvious role, adding that the more versions OpenAI releases, the harder it becomes to tell users which one to pick. The same episode ran the economics: 75 dollars per million input tokens and around 150 per million output tokens, nearly a hundred times the cost of 4o, which rules it out for API-based assistants. By episode 049 one host admitted he had dropped GPT-4.5 because of odd answers, one more argument behind the channel's scepticism about OpenAI's expensive agents.

By the summer of 2025 the verdict had not moved. In episode 061, comparing GPT-4o, o3 Pro and Deep Research, the hosts remarked that OpenAI was trying to position GPT-4.5 as a model for creative writing, something Sam Altman had said once or twice, which did little for the user. In later episodes such as 089 and 133 the model appears only in passing, one more entry in the stream of versions the channel considers the main burden on anyone trying to choose a tool.