GPT-4 Turbo Key Updates

The new GPT-4 Turbo updates have brought plenty of stability and efficiency improvements to the existing GPT 4 model. This model is more capable, cheaper and supports a larger context window. It has an updated knowledge cutoff of April 2023. It introduces a 128k context window (the equivalent of 300 pages of text in a single prompt). The model is also 3X cheaper for input tokens and 2X cheaper for output tokens compared to the original GPT-4 model. The maximum number of output tokens for this model is 4096.

Anyone with an OpenAI API account and existing GPT-4 access can use this model. The most recent version of the model can be accessed by passing gpt-4-turbo as the model name in the API. GPT-4 Turbo also supports DALL-E 3 AI-generated images and text-to-speech.

GPT 4 model price comparison

GPT-4 Turbo provides the following features –

• Updated Knowledge Cut-off: The GPT-4 Turbo model has been updated with a knowledge base up to Apr 2023. Now it can provide more up-to-date information and provide better results.
• More Context Window: Now more information can be provided to the GPT model as input, and its context size has been increased to 128k. 128k context size is equivalent to approx. 300 pages.
• Improved Instruction Following and JSON mode: GPT-4 Turbo performs better than predecessor models on tasks that require the careful following of instructions, such as generating specific formats (e.g., “always respond in XML”). It also supports new JSON mode, which ensures the model will respond with valid JSON. The new API parameter ‘response_format’ enables the model to constrain its output to generate a syntactically correct JSON object.
• Reproducible Outputs and Log Probabilities: The new seed parameter enables reproducible outputs; it makes sure to return consistent output most of the time. This feature is beneficial for debugging, robust unit testing, and having a higher degree of control over the model behaviour. GPT-4 Turbo model also provides log probabilities for generated tokens, this feature could be useful in functionalities like search autocomplete.
• Cost-effective: The GPT-4 Turbo model is cheaper to use. It costs $0.01 for 1,000 prompt tokens and $0.03 for 1,000 completion tokens.

GPT-4 Turbo with Vision: This is a version of GPT-4 Turbo that includes the capability of optical character recognition (OCR). This model can take images as input and the following can be done with this API –

• Generate captions
• Read physical media
• Identify objects

OpenAI also provides a feature where users can create custom chatbots using their platform. Previously, only enterprise and business users could customize ChatGPT for specific industries and use cases. This new feature will enable consumers to encounter personalized AI chatbots on various platforms, including frequently used apps and websites.