OpenAI Unveils New Models and Slashes API Pricing

Here are the main points

  •  OpenAI has introduced embedding models with better performance.
  •  The cost per 1,000 tokens for these models has been significantly reduced.
  •  An updated version of GPT 3.5 Turbo will be launched with pricing.
  •  Improved content moderation capabilities have been integrated.
  •  Developers now have control over API keys and can access usage metrics.

Introduction

OpenAI, the AI lab based in San Francisco has made updates that will greatly benefit developers using their AI services. These updates include the introduction of embedding models a reduction in price for GPT 3.5 Turbo, an updated preview of GPT 4 Turbo and stronger content moderation capabilities.

New Embedding Models

OpenAI has released new text embedding models called text embedding 3 text embedding 3 large. These models offer improved performance compared to generations. For example the text embedding 3 large model achieves a score of 54.9% on the MIRACL benchmark and 64.6% on the MTEB benchmark. This is an improvement, from the older text embedding ada002 model which scored 31.4% and 61% respectively.

Price Reduction

OpenAI has recently made a price reduction specifically for the text embedding 3 small model compared to text embedding ada 002. The cost per 1,000 tokens has dropped by 5 times from $0.0001 to $0.00002. This reduction aims to make these models more affordable for developers. Additionally developers have the option to shorten embeddings in order to reduce costs without affecting accuracy.

Needed Updates

An updated version of GPT 3.5 Turbo is set to be released week along with a 50% decrease in pricing for input tokens and a 25% decrease for output tokens. This marks the time that OpenAI has reduced prices for GPT 3.5 Turbo within the past year showcasing their dedication to fostering wider adoption.

OpenAI has also introduced an update to their GPT 4 Turbo preview with version gpt 4 0125 preview, which has seen over 70% of requests transitioned since its release. This update includes enhancements in code generation and other tasks resulting in comprehensive completions.

New Models

In terms of content moderation OpenAI has launched their advanced model yet; text moderation 007. This model excels at identifying potentially harmful text compared to previous versions making it an invaluable tool, for developers focused on creating safer AI applications.

Improved Management of API Keys

Developers now have enhanced control over their API keys. Can gain insights into usage metrics. According to OpenAI developers have the ability to assign permissions to individual keys and track consumption on a per key basis. This enables them to effectively monitor the performance of their products or projects.

In Conclusion

OpenAI isn’t stopping at these updates alone. The AI research lab has plans for enhancements to its platform in the upcoming months specifically targeting larger development teams. By improving its models and offering price reductions OpenAI demonstrates a clear commitment, to advancing the AI industry as a whole.

Would you like to learn more about how Chapin Industries Group may be able to help you? You can use the form to submit a request for FREE consultation!