Despite the release, the Grok-1 model requires substantial work to operate in a conversational context, and its large size may limit its use to datacenter-class inference hardware. The model has been released under the Apache 2.0 license, and the inference code and weights are available for download. However, the industry has not yet agreed on a term for AI model releases that ship code and weights without also releasing training data, so these releases are typically referred to as "source available" or "open weights".
Key takeaways:
- Elon Musk's AI firm xAI has released the base model weights and network architecture of Grok-1, a large language model designed to compete with OpenAI's ChatGPT.
- The release comes amid Musk's criticism and legal action against OpenAI for not releasing its AI models in an open way.
- The base model of Grok-1 is not fine-tuned for a specific task and requires substantial extra work to operate in a conversational context.
- xAI has released the base model weights and network architecture under the Apache 2.0 license, but due to its size, it requires datacenter-class inference hardware to run.