ChatGPT offers exceptional support for creative writing tasks. The platform’s newest large language model, GPT-4o, has recently been enhanced, as OpenAI shared on Wednesday. Users can anticipate “more natural, engaging, and personalized writing that improves relevance and readability” in future interactions.
GPT-4o received an upgrade! 🎉
The model’s creative writing capabilities have significantly improved—delivering more natural, engaging, and tailored content that enhances relevance and readability.
It also excels at working with uploaded files, yielding deeper insights and more comprehensive responses.
— OpenAI (@OpenAI) November 20, 2024
It’s important to note that GPT-4o is not to be confused with the earlier 01 model (previously known as Project Strawberry). Released in May 2024, GPT-4o is OpenAI’s most current public offering, outpacing both GPT-4 and GPT-3.5. This model delivers double the performance at half the resource expenditure compared to its closest predecessor, GPT-4-Turbo, and showcases impressive results across benchmarks in voice, multilingual, and visual tasks. Alongside its efficiency, GPT-4o also introduces numerous additional features, making it particularly advantageous for applications requiring real-time translation and conversational interactions.
For instance, the Advanced Voice Mode in ChatGPT is made possible by the near-instantaneous text, voice, and audio processing that GPT-4o provides, a significant enhancement compared to earlier models. The new version excels in reasoning capabilities, enabling it to detect a user’s tone and intent, interpret distinctive aspects of their speech, and respond accordingly.
While GPT-4o is accessible to all OpenAI subscribers, including those on the free tier, usage is not unlimited. Users on the complimentary plan can engage with the upgraded model only a limited number of times in ChatGPT before switching to the smaller GPT-4o-mini model. Teams and Enterprise users benefit from a usage capacity that is approximately five times greater.
GPT-4o-mini utilizes the same training data as its larger counterpart but operates with fewer variables and lower resource requirements. Thanks to its streamlined design, GPT-4o-mini proves effective in several small-scale applications, including code generation.
According to OpenAI, GPT-4o mini is touted as “the most capable and cost-effective small model available today.” As reported by CNBC, it outperforms rivals such as Google’s Gemini 1.5 Flash, Meta’s Llama 3 8b, and Anthropic’s Claude 3 Haiku across various benchmarks. Data from Artificial Analysis indicates that GPT-4o-mini achieved an 82% score on the MMLU reasoning benchmark, surpassing Gemini by 3% and Claude 3 Haiku by 7%.
