In a livestream today, Openai finally announced the introduction of his GPT 4.5 model, but with a turn: for the moment it requires a Chatgpt Pro subscription of $ 200 per month.
This is because the new major language model (LLM) is still technically in a “research preview” status. This is apparently short form for “Please pay us for the privilege of being beta tester.”
Apart from jokes, the company promises GPT-4.5 “natural conversation” and works better in terms of programming, pattern recognition, writing and “practical problems”.
Note that GPT-4.5 is not intended as an argumentation model, which means deep research Skills of Openai O1Present O3 miniOr the upcoming full version of O3. In A leaked documentThe company said Is Openai’s largest LLM that should make it ideal for everyday questions because the large database with training material should provide faster answers.
Credit: Openai
In internal tests, Openaai says the new model hallucinated About 24.7% less often than GPT-4.0 and were approximately 34.3% more precisely. Around 57% of internal human testers preferred GPT-4.5 to 4.0.
Credit: Openai
In other words, GPT-4.5 is not up to date with the latest status of the AI, but it should be an improvement for other non-renowned models and offer a nice middle ground between cheaper, more traditional LLMs and their often pawaled successors. The only larger versions seem to be multimodal functions such as AI language mode and video inputs, although Openaai, according to Openai, to update his user experience, which apparently could result in a version of these functions to the model in the future.
This makes sense because this is once again an early publication, with Openaai the boundaries of GPT-4.5 and how humans can use it. For this purpose, the LLM will be expanded to all paid chatt plans in the next two weeks before you have probably reached free users as a replacement for GPT-4.0 as soon as it leaves its preview.
