GPT-4 Turbo

What is OpenAI GPT-4 Turbo?

GPT-4 Turbo is a version of OpenAI's GPT-4 model, announced in November 2023. It has a knowledge cutoff of April 2023, compared to September 2021 for the original GPT-4, and an expanded context window of 128k tokens (equivalent to over 300 pages of text), allowing it to process longer documents and conversations in a single prompt.

In benchmarks conducted by Klu.ai, it was observed that when dealing with 60-128k input tokens, GPT-4 Turbo tends to more consistently retrieve facts from the latter 50% of the input.

Benchmarks tested with randomly sampled facts with no intermediate libraries (LangChain, LlamaIndex, etc) or context systems (including Klu Context).

Key features of GPT-4 Turbo compared to the original GPT-4 include:

• Updated knowledge base — Trained with information up to April 2023, giving it more current context in its responses.
• Larger context window — A 128k token context window, versus GPT-4's 8,192 tokens.
• Lower cost — Priced at $0.01 per 1,000 input tokens and $0.03 per 1,000 output tokens, roughly a third of the cost of GPT-4 for input.
• Multimodal capabilities — Supports image inputs, along with DALL-E 3 image generation and text-to-speech with several preset voices.
• Customizable chatbots — OpenAI's GPTs feature allows users to create custom versions of ChatGPT for specific purposes.
• Accessibility — Available via the API by passing a specific model name, such as gpt-4-1106-preview.

How can developers access GPT-4 Turbo?

Developers need an OpenAI API account with existing GPT-4 access. The model can be accessed by passing gpt-4-1106-preview as the model name in the API.

Here's a Python example of how to make a request to the GPT-4 Turbo API:

from openai import OpenAI

client = OpenAI()
completion = client.chat.completions.create(
    model="gpt-4-1106-preview",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
    ]
)