W&B Models

What it does

W&B Models takes in model experiment data, including metrics, multimodal experiment runs, and large model files, and produces tracked experiment records, versioned artifacts, comparisons, and visualizations. Teams use it to log training runs, compare results across experiments, and identify improvements or regressions before moving models toward production.

Its distinct focus is experiment tracking and training observability. The page emphasizes low-latency, high-throughput logging for large numbers of experiments and metrics, and deployment flexibility across managed cloud, private cloud, on-premises, and hybrid setups. It also supports bringing your own storage bucket for regulated or security-sensitive data. It is separate from inference hosting and Serverless RL products.

W&B Models suits ML teams that train or fine-tune models and need reproducible experiments, collaboration and access controls. Forge provides free personal-development access and paid team plans; private deployment terms are separate. Storage allowances and overage affect the cost of retaining datasets and model artifacts. Confirm security features and deployment terms for the selected plan, and avoid logging secrets or data your project is not authorized to retain.

What it can do

• Track and version multimodal model experiments
• Compare experiment runs and visualize performance metrics
• Store and version datasets and model artifacts
• Log and monitor large numbers of experiments and metrics at scale
• Build collaborative dashboards and reports
• Manage access controls, service accounts, and audit logs

Example workflows

Track a model training run

Illustrative example. Input: a model training job with metrics and checkpoints. Steps: log experiment metrics and artifacts to W&B Models, then open the run in the dashboard. Output: a tracked run with visualized metrics and versioned model files.

Compare experiment recipes

Illustrative example. Input: multiple fine-tuning or dataset variants logged as runs. Steps: select the runs in the project and compare them side by side. Output: a comparison view of performance differences for identifying improvements or regressions.

Before you choose

• Storage quotas and overage affect retained datasets and model artifacts.
• Forge Pro eligibility is limited to early-stage teams with fewer than 50 employees; larger organizations need Enterprise terms.
• Check the supported deployment and security controls for your plan.

Listed under

• Coding
• AI training
• Experiment tracking
• Hyperparameter search
• ML engineering
• MLOps
• ML teams

Get started

Create an account to get started; a 30-day free trial is offered on the Pro plan.

1. Create an account at the sign-up page.
2. Log in and create a project to hold experiments.
3. Install the open-source SDKs from the GitHub library linked on the product page and begin logging runs.

Forge Free is for personal development. Pro starts at $60/month billed monthly after a 30-day trial and is limited to eligible early-stage teams. Model artifact storage has separate allowances and overage; Weave data-ingestion charges are for the separate Weave product.

30-day free trial.

Source: coreweave.com/forge-pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Weights & Biases Models | Track, compare, improve models ↗www.coreweave.com
• CoreWeave Forge Pricing ↗www.coreweave.com

Company website

coreweave.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.