After yesterday’s announcement, ChatGPT-4.5 will be rolling out today and over the next couple of days, I was very pleased to see it appear in my model selector. However, the model selector looks like a complete muddle now.
Have a look at:
There are eight different options. OpenAI has provided handy explanations for each option, such as “Great for most questions”or “Uses advanced reasoning”but it’s still a long list to understand.
OpenAI understands
OpenAI is aware that this is not a good solution. OpenAI CEO Sam Altman announced in February that they would simplify their model selection to make it more consumer-friendly. But now ChatGPT 4.5 is here and the situation is worse than ever. Altman posted on X in February a post entitled “OPENAI ROADMAP UPDATE FOR GPT-4.5 and GPT-5″in which he said, “We want AI to ‘just work’ for you; we realize how complicated our model and product offerings have gotten. We hate the model picker as much as you do and want to return to magic unified intelligence.”
Fair enough, Altman did mention that after releasing ChatGPT 4.5, “a top goal for us is to unify o-series models and GPT-series models by creating systems that can use all our tools, know when to think for a long time or not, and generally be useful for a very wide range of tasks.”
Unfortunately there is no timeline for when that will happen, and the current LLM picker screen (Large Language Models) is beyond ridiculous. ChatGPT users who are on the free plan have a much simpler experience. There is only one option, “ChatGPT”and an option to upgrade to the Plus plan.
ChatGPT 4.5 works well
From my initial testing, ChatGPT 4.5 worked flawlessly. It feels like a mix of OpenAI’s deeper reason models such as o1 and O3 with a little bit of ChatGPT-4o’s casual chat. It’s also fast, which is great news for all ChatGPT Plus subscribers who have access to the LLM. They can test it right now. OpenAI describes ChatGPT 4.5 as feeling more natural. “It’s broader knowledge base, improved ability to follow user intent, and greater ‘EQ’ make it useful for tasks like improving writing, programming, and solving practical problems. We also expect it to hallucinate less.”
The problem of hallucinating or “making stuff up” to the rest the world, has been a major issue for most chatbots. If OpenAI can make significant improvement in this area, we will all be grateful. Over the next few days, we’ll give you a more thorough assessment of what ChatGPT 4.5 can do.
- Deep Research is like a super-smart but slightly absentminded librarian in a children’s story
- ChatGPT can give me answers, but Deep Research gives a dissertation I will never use
- With ChatGPT Voice mode now free, does it make sense to pay for ChatGPT Plus?
Graham, Senior Editor at TechRadar for AI. Life, Maximum PC, and more. He has been featured on BBC TV shows such as BBC One Breakfast, and Radio 4 to comment on the latest tech trends. Graham holds an honors degree from the Department of Computer Science. He spends his free time podcasting and blog.
Latest in Artificial Intelligence (19659022)
More on artificial intelligence
