OpenAI Showcases New Generative AI Models and Lower API Prices, Cures GPT-4 ‘Laziness’
OpenAI has debuted new variants of accessory generative AI models and upgraded its GPT-4 Turbo and GPT-3.5 Turbo models, slashing the latter’s API prices enormously. The enhancements include making GPT-4 Turbo better at fully completing prompts, instead of being ‘lazy’ and only partially carrying out user requests.
Better, Cheaper LLMs
The pricing shift accompanying the updated GPT-3.5 Turbo, which powers the free version of ChatGPT and most generative AI tools businesses are licensing OpenAI’s LLMs, jumps out as the biggest aspect of the update, especially as it is the third in less than a year. OpenAI halved the input cost to $0.0005 per 1,000 tokens and took 25% off the output price to make it $0.0015 per 1,000 tokens. The actual updates to the model are left unspecified, however. The improvements to GPT-4 Turbo are also left vague except for an update trying to address the issue of the model only partly completing a task.
“Over 70% of requests from GPT-4 API customers have transitioned to GPT-4 Turbo since its release, as developers take advantage of its updated knowledge cutoff, larger 128k context windows, and lower prices,” OpenAI explained in a blog post. “Today, we are releasing an updated GPT-4 Turbo preview model, gpt-4-0125-preview. This model completes tasks like code generation more thoroughly than the previous preview model and is intended to reduce cases of “laziness” where the model doesn’t complete a task.”
OpenAI also rolled out two new text embedding models for developers. Embeddings are mathematical expressions of words and concepts that an AI model can use to interpret conceptual relationships. The release includes both a smaller, more efficient model and a larger, more powerful option. Both models apply an advanced training technique allowing developers to shorten the embeddings for more efficient storage without sacrificing too much accuracy. This enables flexible implementation in apps and databases. OpenAI also upgraded its content moderation model that screens for harmful text as part of its safety efforts. There are also new controls to allow users to better manage API keys and monitor detailed usage metrics for billing purposes.
Follow @voicebotaiFollow @erichschwartz
OpenAI Enhances GPT-4 and GPT-3.5 Knowledge and Memory, Drops Prices and Ups Legal Protection
OpenAI Upgrades GPT-4 and GPT-3.5 Turbo Models, Reduces API Prices
