Nothing. Google has PALM, Meta has FLAVA, Amazon has AlexaTM/CoT, every major tech company has been working on LLM and multimodal models for years. OpenAI was always the benchmark, but the major tech companies are not that far behind - maybe 2-3 years at most. The parameters race has been escalating for awhile now at a rapid pace, so these models were always going to get to this level soon once they scaled sufficiently and had proper training data.

What GPT did better than anyone, was make it a Chatbot. It showcased the power of AI to normal people who don't understand ML. They're also much better at cleaning their data and scaling their model than companies who are more focused on specific business use cases than generic knowledge models.