Earlier this week, OpenAI showcased its spring updates. While some were speculating on a GPT-5 announcement, the update ultimately did not include a new foundational model but was largely associated with improvements to the ChatGPT product and the GPT-4 foundational model.
Nonetheless, the latest iteration of the GPT-4 model, dubbed GPT-4o, provides a notable step up in performance, accessibility, and multi-modality usage (i.e., text, audio, vision):
- The system can respond to audio prompts within 200-300 milliseconds. OpenAI claims it’s “2x” faster than previous versions.
- OpenAI is lowering the cost to access it by half and increasing the rate limits by 5x relative to GPT-4 Turbo.
- The ChatGPT product will integrate with the text and image functionality, and GPT-4o will be available for free-tier users.
In addition to the spring updates, it appears increasingly likely that OpenAI’s ChatGPT will soon be integrated with Apple’s iOS.
OpenAI’s focus on increasing the underlying model capabilities by layering on performant multi-modal functionality and seeking greater distribution via Apple suggests that it’s likely leaning into creating a consumer-focused product.
Blockworks Intel