OpenAI Unveils GPT-5.4 with Enhanced Capabilities
- GPT-5.4 features a context window of up to 1 million tokens, surpassing previous models.
- The model scored 83% on the GDPval test, improving from 70.9% with GPT-5.2.
- Two versions are available for different needs: GPT-5.4 Pro and GPT-5.4 Thinking.
- Error rates have decreased by 33% for factual inaccuracies and 18% overall.
- A new system evaluates reasoning chains to enhance transparency during problem-solving.
The launch of GPT-5.4 signifies a major advancement in AI technology, particularly beneficial for sectors requiring extensive data analysis, including cryptocurrency markets where accuracy is crucial.
With an impressive score of 83%, GPT-5.4 demonstrates significant improvements in performance, reflecting OpenAI’s commitment to enhancing AI capabilities across various industries.
