OpenAI released GPT-4.5, but it's not much of an upgrade from GPT-4o
Don't expect to be surprised.
2 min. read
Published on
Share this article
Read our disclosure page to find out how can you help Windows Report sustain the editorial team. Read more
After DeepSeek was unleashed into the world, everyone wondered what OpenAI would do now that another AI company had developed an extremely powerful model at a tiny fraction of the budget.
Well, the answer is GPT-4.5, which was teased a few weeks ago. Even though OpenAI says it’s their biggest model yet, it’s not much of an upgrade compared to GPT-4o. It’s not even doing that well compared to the other models on the market.
According to OpenAI’s blog post, there are the overall stats of GPT 4.5 compared to GPT-4o:
| GPT‑4.5 | GPT‑4o | OpenAI o3‑mini (high) | |
| GPQA (science) | 71.4% | 53.6% | 79.7% |
| AIME ‘24 (math) | 36.7% | 9.3% | 87.3% |
| MMMLU (multilingual) | 85.1% | 81.5% | 81.1% |
| MMMU (multimodal) | 74.4% | 69.1% | – |
| SWE-Lancer Diamond (coding)* | 32.6% $186,125 | 23.3% $138,750 | 10.8% $89,625 |
| SWE-Bench Verified (coding)* | 38.0% | 30.7% | 61.0% |
As you can see, the overall score between the models is almost similar in many instances.
OpenAI says the new model has stronger reasoning and doesn’t think before it responds, making it a more general-purpose model. It’s apparently trained for human collaboration and will better understand nuance and natural conversation than the previous models.
It will also hallucinate less than other GPTs; however, the model is has scored medium to low points in security.
OpenAI says GPT-4.5 is now available in preview for ChatGPT Pro users on web, mobile, and desktop, so if you already have a subscription, try it.
But don’t expect to be surprised.
User forum
0 messages