Weights & Biases

Weights & Biases is used by more than 1 million AI practitioners at 1,400+ companies including OpenAI, Meta and NVIDIA, with more than 1 billion tracked runs; it is now integrated into CoreWeave Forge for model and agent development from training through evaluation and production.

Weights & Biases reports 900,000 users and 1,000+ companies, including OpenAI, NVIDIA, Canva, Toyota Research Institute, Salesforce, Microsoft and other major AI teams.

Weights & Biases reports 900,000 users and 1,000+ companies, including OpenAI, NVIDIA, Canva, Toyota Research Institute, Salesforce, Microsoft and other major AI teams.

Weights & Biases has evolved into an end-to-end AI developer platform spanning model experiments, sweeps, registry and artifacts plus Weave traces, agent evaluations, guardrails and monitors; CoreWeave ARIA adds an AI research agent that analyzes experiment data and production traces and can design and launch follow-up experiments.

Weights & Biases provides experiment tracking, model development, evaluation, observability and Weave tooling across ML and generative-AI workflows.

Weights & Biases provides experiment tracking, model development, evaluation, observability and Weave tooling across ML and generative-AI workflows.

Weights & Biases was acquired by CoreWeave in May 2025, integrating its developer platform with a major AI cloud provider and materially increasing compute, capital and organizational resources.

Weights & Biases was acquired by CoreWeave in May 2025, integrating its developer platform with a major AI cloud provider and materially increasing compute, capital and organizational resources.

CoreWeave completed its acquisition of Weights & Biases in May 2025, combining W&B with a public AI-cloud company and subsequently integrating it as a named product within CoreWeave Forge while retaining its platform, customers and cloud-agnostic operation.

Weights & Biases connects experiment tracking, model evaluation and model management with CoreWeave Forge Training, Inference, Registry and Sandboxes while remaining cloud-agnostic, supporting portability across ML research, evaluation, MLOps, inference and AI-platform engineering.

Experiment tracking, evaluation, observability, model development and AI platform skills transfer exceptionally broadly across frontier labs, cloud and enterprise AI.

Experiment tracking, evaluation, observability, model development and AI platform skills transfer exceptionally broadly across frontier labs, cloud and enterprise AI.