Weights & Biases (often abbreviated W&B) is an AI developer platform widely used for experiment tracking, model and dataset versioning, collaboration, and large language model evaluation and observability. Founded in 2018 and headquartered in the United States, the company became a standard tool in many machine learning teams' workflows through a lightweight SDK that integrates with major ML frameworks.
The core platform lets teams log training metrics, visualize model performance, compare experiments, and manage model artifacts so that machine learning work is reproducible and auditable. As generative AI adoption grew, W&B added Weave, a toolset focused on tracing, evaluating, and monitoring large language model applications, extending the platform from traditional model training into LLM development and production observability.
W&B has been used by AI teams at major organizations and research labs, and the company reported a large community of developers relying on its tooling for model development and evaluation. Its appeal centers on reducing the bookkeeping burden of ML experimentation while improving collaboration across research and engineering teams.
In 2025, CoreWeave completed its acquisition of Weights & Biases, with reporting describing the transaction at around $1.7 billion. Following the acquisition, W&B operates as part of CoreWeave's AI cloud platform, combining experiment tracking and evaluation tooling with CoreWeave's compute infrastructure. CoreWeave has publicly emphasized continued interoperability so customers can keep using multiple infrastructure providers, frameworks, and models.
W&B competes with other MLOps and experiment-tracking tools as well as LLM observability and evaluation platforms. Its differentiation has rested on broad framework integration, mature visualization and collaboration features, and now its position within a large AI cloud provider.
The platform is best suited to machine learning and AI engineering teams that need rigorous experiment tracking, model governance, and LLM evaluation, particularly those comfortable operating within or alongside the CoreWeave ecosystem post-acquisition.