OpenAI Upgrades ChatGPT Premium With Smarter GPT-4 Turbo Model
OpenAI releases an improved GPT-4 Turbo model for its paid ChatGPT tiers, offering better performance in writing, math, and coding. The update extends the model's knowledge base to December 2023 and shows significant accuracy gains on complex benchmarks.
OpenAI releases an upgraded version of its GPT-4 Turbo large language model to paid ChatGPT users. The company states that this new iteration improves capabilities in writing, math, logical reasoning, and coding, building upon the foundation of the previous GPT-4 Turbo model that launched six months ago.
The latest model features an expanded knowledge base that includes information up to December 2023, whereas the previous version only accesses data from before April 2023. Using an internally developed benchmarking tool called Simple Evals, OpenAI tests the new model across six different question sets and finds that it outperforms its predecessor in every category.
The most notable accuracy gains occur on the GPQA benchmark, which contains graduate-level multiple-choice questions spanning physics and math, where the model achieves a nearly 10% improvement. This updated GPT-4 Turbo model is currently available to subscribers of the ChatGPT Plus, Team, and Enterprise tiers.