Microsoft has secretly included the most recent iteration of OpenAI’s massive language model GPT-4-Turbo into its Copilot AI helpers. The previous iteration of the model was exclusively accessible with a $20 monthly premium subscription to ChatGPT Plus. It enables more intricate questions and answers. These days, Copilot may be found in a variety of Microsoft products and services, including Windows 11, the 365 productivity suite, and stand-alone apps for iOS and Android. GPT-4-Turbo is gradually being made available to Copilot customers who have activated GPT-4. It offers a more recent knowledge cutoff, more processing power, and more memory.

What distinguishes GPT-4-Turbo?
The extensive language models that support both Microsoft’s and OpenAI’s AI services and products are being updated regularly. This covers improving and fine-tuning skills.

Compared to previous iterations, the most recent version of GPT-4, which is also its most sophisticated model, features an updated knowledge cutoff. It includes data until April 2023 and a significantly bigger 128k context window, which is the equivalent of 300 text pages in memory.

One year prior to ChatGPT’s introduction, in September 2021, the base GPT-4 model’s previous data cutoff was reached. Additionally, after reading through about 100 pages of text, it would forget what you had told it because of its 32,000 token context window.

Another noteworthy modification in GPT-4-Turbo is its enhanced picture analysis and reasoning skills.

It is therefore more adept at adhering to directions. According to a blog post by OpenAI, “GPT-4 Turbo performs better than our previous models on tasks that require careful following of instructions”.

Although GPT-4-Turbo isn’t officially available within Copilot, Microsoft did state in December that the AI tool will receive frequent updates.

Some users observed that the model named after the most recent OpenAI LLM was listed in the Copilot website’s source code.

Topics #ChatGPT #Microsoft