OpenAI's GPT-4.5: Larger Model, Mixed Results
Tap title to read story
OpenAI launched GPT-4.5, its largest model yet, available to ChatGPT Pro subscribers and API developers. While trained with more data and computing power than previous models, its performance shows diminishing returns from scaling. GPT-4.5 outperforms predecessors on some benchmarks, especially in areas like emotional intelligence and creative tasks, but lags behind newer reasoning models from competitors on others. OpenAI views it as a research preview and stepping stone towards more powerful future models, acknowledging limitations in the scaling approach.
JQJO Editorial Desk · Human reviewed · AI assisted. Report a correction.