OpenAI recently released its newest GPT-4o version. GPT-4o (“o” for “omni”) is a multimodal GPT model that is similar to GPT-4 Vision and Turbo but is much faster and is also trained on the data up to October 2023. OpenAI also significantly reduced costs for GPT-4o tokens, so we decided to pass on these savings to our +Drafting subscribers to provide you with complimentary access to GPT-4o within ClaimMaster’s patent drafting tools (we might revisit this later depending on long-term usage/costs). This will give you access to the latest, most powerful generative AI with vision capabilities to be used for various patent drafting tasks within ClaimMaster.
We’ve set up two complimentary GPT-4o models, one hosted in OpenAI cloud and one hosted in Microsoft Azure cloud. Both of these models are stateless and are NOT getting trained on any data/prompts you submit to them. Specifically, when connecting to the cloud-based GPT services, ClaimMaster will send and receive data to/from the selected GPT service using Enterprise API directly from your computer. According to OpenAI API policies and/or Azure OpenAI API policies, the use of Enterprise API ensures that your prompts (inputs), completions (outputs), and any other transmitted data will not be available to other customers and will not be used to improve their GPT models. Moreover, all communications with the selected GPT service will be directly from your computer without involving any other 3rd party servers for data processing.
To enable access to GPT-4o in ClaimMaster for +Drafting subscribers (i.e., Pro+Drafting or Pro+Shells+Drafting), perform the following steps:
- Download and install the latest ClaimMaster release (2024.1.1) from the Customer Portal.
- Simply set the GPT Source to be either OpenAI GPT or Microsoft Azure GPT in ClaimMaster’s GPT/LLM settings, while leave the rest of the fields/settings blank. All subsequent GPT prompts will go to the configured GPT source and will internally use the latest GTP-4o model for generating text without paying extra for your +Drafting subscription.
For those who already have or would like to set up their own private models with OpenAI or Microsoft Azure, you can configure ClaimMaster to use those private models instead.
Overall, GPT-4o has excellent vision capabilities that make it well-suited for analysis of figures when drafting patent applications. It could be a real time saver! Here are a few things that +Drafting subscribers can do with the new models:
- Pass in figures and have GPT write first drafts of detailed figure descriptions. Here’s more information on the specific drafting prompts that may be helpful for writing the detailed description section.
- You can also pass GPT schematics (either CAD generated or sketches) and ask it to generate a list of all part names and numbers in each figure.
- As with previous models, you can as GPT to generate various patent document sections for you and also use it as a powerful re-writing assistant.
In any event, if you are a +Drafting subscriber, you now have access to the latest GPT-4o models and can test them out for variety of patent drafting tasks. Enjoy, it’s on the house!