IBM is excited to offer a 30-day demo, in which you can chat with a solo model to experience working with generative AI in the IBM^® watsonx.ai^™ studio.  

In the watsonx.ai demo, you can access some of our most popular AI models, ask them questions and see how they respond. This gives users a taste of some of the capabilities of large language models (LLMs). AI developers may also use this interface as an introduction to building more advanced AI solutions.

What can you do with the watsonx.ai demo?

In this first release of the watsonx.ai demo, you can access 5 model types:

• granite-13b-chat-v2, trained by IBM Research on enterprise-relevant data
• llama-3-8b-instruct, new model, optimized for speed
• llama-3-70b-instruct, new model, optimized for accuracy
• llama-2-13b-chat, optimized for speed
• llama-2-70b-chat, optimized for accuracy

You can start multiple chat sessions with any of these models and use them for various tasks, including providing information and research help, engaging in conversation, aiding in education, offering creative assistance, problem solving, and entertainment through games and quizzes. Although the tool is not designed as a consumer solution, it can be a valuable resource for AI developers and researchers to test some watsonx AI models.

As part of your watsonx.ai demo, you will have access for 30 days and receive 20,000 free tokens. If you run out of tokens, you can subscribe to watsonx.ai.

Simplifying the AI development workflow

With our watsonx.ai demo, you can interact with a solo LLM. This demo does not include agents, simultaneous chat with multiple models, multi-modal models, or other functionality to enhance results. Models might not have knowledge of recent events.

During the AI solution development lifecycle, it is crucial to quickly test multiple models, interact with them, collect feedback and share them with peers. With this new demo tool, we aim to pave the way for a future of accelerated time to value for popular common patterns.

In the future[1], product developments are planned to allow watsonx.ai studio users to chat with fine-tuned versions of their models. They may continue the customization workflow in our suite of development tools that the full watsonx.ai studio offers, such as Prompt Lab and Tuning Studio, or by using our APIs or Python SDK.

This customization may involve:

• Building a more advanced RAG solution
• Tuning the model by using multiple techniques
• Optimizing the model to reduce production costs

Expanding capabilities of our Prompt Lab with Chat mode

Last year, we released Prompt Lab, a tool in IBM watsonx.ai for experimenting with prompting different foundation models, exploring sample prompts, and saving and sharing your best prompts. Users can select any model from this tool and try different prompting techniques, such as zero-shot, one-shot, few-shot and chain-of-thought. Prompts may be saved as reusable assets to build end-to-end AI solutions.

The Prompt Lab on watsonx.ai provides advanced options for selecting the model parameters and variables. For example, you may customize a model’s decoding techniques to choose the tokens in the generated output or add AI Guardrails to help filter harmful language from the input prompt text and the output generated by the model.

Now, in the Prompt Lab, you have a new view to interact with models called Chat mode, which adds to the existing views, Freeform and Structure. These multiple options will allow you to craft the model configurations.

By using the Chat mode in the Prompt Lab through our free trial or paid plans, you can see the benefits of chatting and building prompts with a variety of foundation models. You may also test them for accuracy and performance before adding them into your end-business applications or AI solutions by using REST APIs.

Try out the new 30-day demo of watsonx.ai today to chat directly with an LLM or access the full watsonx.ai trial to learn how to find what you need to prepare data and build models, experiment with foundation models and deploy your models.

[1] Statements regarding IBM’s future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.