Weights & Biases
Weights & Biases is an AI developer platform for tracking experiments, evaluating LLM apps and agents with Weave, and managing models; it is now part of CoreWeave Forge.
About this tool
Weights & Biases, often shortened to W&B, is a long-standing platform for machine learning teams. Adding a few lines of code to a training script logs metrics, hyperparameters and outputs, so runs can be compared in dashboards and reproduced later. Its Weave product applies the same idea to generative AI, tracing and evaluating LLM applications and agents.
Weights & Biases now belongs to CoreWeave, and its site presents it as part of CoreWeave Forge, a product family for developing, evaluating and serving models and agents. Model training, inference and serving capabilities have moved under the Forge name. W&B remains free for academic research, offers an iOS app to monitor runs, and lists several compliance certifications.
What you can do
Experiment tracking
Log metrics, configurations and artefacts from training runs and compare them in shared dashboards.
Weave for LLM apps
Trace, evaluate and improve generative AI applications and agents by adding a library to your code.
Model lifecycle management
Register and manage models and prototype in notebooks across the development lifecycle.
Mobile monitoring
An iOS app lets teams follow experiments and training runs away from their desks.
Who is it for?
- A research lab compares hundreds of training runs to choose the best hyperparameters.
- A product team evaluates prompt changes to an LLM feature before shipping them.
- A university group tracks thesis experiments under the free academic licence.
Before you choose
Product names, plans and documentation are moving under CoreWeave Forge, so check current naming and pricing before procurement.
Compliance certifications are stated by the vendor and depend on the deployment option; request the relevant reports for your review.
Frequently asked questions
Is Weights & Biases still available now that it is part of CoreWeave?
Yes. The site presents Weights & Biases as part of CoreWeave Forge, and its tracking and Weave products remain available.
Is Weights & Biases free?
It is free to start building and free for academic research. Paid usage depends on factors such as storage and Weave data ingestion.
What is W&B Weave?
Weave is the Weights & Biases toolkit for tracing and evaluating LLM applications and agents, added to an app with a library.
Check the official website for current features, pricing, and terms.
Similar tools
Mistral AI
CodingMistral AI is a French AI company offering open-weight and frontier models, the Studio developer platform and API, Forge for customisation and its Vibe assistant.
LangChain
CodingLangChain provides open-source frameworks for building AI agents and LangSmith, a platform to trace, evaluate, deploy and monitor them in production.
Lunary
OtherLunary is an open‑source platform that helps developers monitor, debug, and improve large language model applications.
DeepSeek
CodingDeepSeek develops large language models, including open-weight reasoning models, offered through a chat app and an OpenAI-compatible developer API.
Modal
CodingModal is a serverless cloud for AI workloads: run inference, batch jobs, fine-tuning, sandboxes and GPU notebooks from Python code, billed per second.
LLM Gateway
OtherLLM Gateway provides developers with a central hub to route, manage, and monitor large language model requests across multiple providers, while tracking usage costs and performance metrics.