OpenAI has introduced GPT-4 Turbo, an enhanced version of its text-generating AI model, GPT-4, which is not only “more powerful” but also more cost-effective. This development was unveiled at OpenAI’s first-ever developer conference. GPT-4 Turbo offers two variants: one dedicated to text analysis and another that comprehends text and images. The text-analyzing model is available for preview via an API, with plans for general availability shortly.
OpenAI has priced GPT-4 Turbo at $0.01 per 1,000 input tokens (approximately 750 words) and $0.03 per 1,000 output tokens. The pricing for the image-processing version depends on the image size. For instance, processing an image with 1080×1080 pixels costs $0.00765. These prices represent a significant reduction compared to GPT-4, with a 3x decrease for input tokens and a 2x decrease for output tokens.
GPT-4 Turbo has several improvements over GPT-4, including an updated knowledge base, extending its understanding to events up to April 2023. This ensures that questions related to recent events are answered more accurately, at least until the new cut-off date.
One notable feature of GPT-4 Turbo is its expanded context window, offering a 128,000-token context window, four times the size of GPT-4’s. This larger context window enhances the model’s ability to maintain context during longer interactions.
OpenAI has also introduced a “JSON mode” in GPT-4 Turbo, which ensures that the model responds with valid JSON, making it more compatible with web applications. Additionally, new parameters allow developers to improve the model’s response consistency and access log probabilities for the most likely output tokens.
OpenAI is launching an experimental access program for fine-tuning GPT-4, with more oversight and guidance from OpenAI teams compared to GPT-3.5. The company has also doubled the tokens-per-minute rate limit for paying GPT-4 customers while maintaining the pricing at $0.03 per input token and $0.06 per output token.
These developments mark a significant step forward in AI text generation, offering more power and capabilities at a reduced cost. GPT-4 Turbo is poised to have a substantial impact on various applications and industries.