Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Weights & Biases

wandb.ai

  • Free plan
  • Windows
  • macOS
  • Linux
  • Web
  • iPhone and iPad
  • Android
  • API
  • Self-hosted
8.6 Club score Visit wandb.ai

What Weights & Biases is

Weights & Biases is an AI developer platform for creating agents, applications, and machine-learning models. Founded in 2017 and headquartered in San Francisco, it combines experiment management, application observability, and asset governance. W&B Models records and visualizes training experiments and offers Sweeps for hyperparameter optimization. Weave traces AI applications and agents, runs evaluations, and monitors their behavior. Registry provides a controlled place to store and version models and datasets, with aliases, lineage tracking, governance, and access controls. The platform also lists model and data versioning, artifact storage, pipeline orchestration, model serving, managed compute, deployment tracking, and model aliases. Integrations named by the maker include cloud providers, OpenAI, Meta, Amazon Bedrock, Anthropic, LangChain, and LlamaIndex. Deployment can use SaaS, dedicated, or customer-managed arrangements. The platform is available through its API, web, Windows, macOS, Linux, iOS, and self-hosted options. The maker describes use cases such as model training, fine-tuning, AI application development, and production monitoring. Security certifications and compliance claims include ISO/IEC 27001:2022, ISO/IEC 27017:2015, ISO/IEC 27018:2019, SOC 2, and HIPAA. An iOS app monitors experiments and training runs.

Who it suits

Weights & Biases suits AI developers and teams working on model training, fine-tuning, agent or application development, and production monitoring. It also fits organizations that need model and dataset versioning, lineage, governance, deployment choices, and integrations with major AI services.

Good to know, and what to watch

  • Tracks and visualizes machine-learning experiments
  • Sweeps supports hyperparameter optimization
  • Weave traces, evaluates, and monitors AI applications
  • Registry versions models and datasets with lineage
  • SaaS, dedicated, and customer-managed deployment options
  • Integrates with OpenAI, Anthropic, LangChain, and LlamaIndex
  • Free plan limits storage to 5 GB per month
  • Free plan allows five model seats
  • Pro is intended for teams under 50 employees
  • Dedicated and on-premises deployments are not supported in the iOS listing

Advanced Enterprise

—

Enterprise

—

Free

Free

Personal

Free

Free plan · 30-day trial

Getting started

  1. Open https://wandb.ai/site
  2. Choose the Free, Personal, Pro, Enterprise, or Advanced Enterprise plan
  3. Use the API, web app, desktop platforms, iOS app, or self-hosted option
  4. Connect listed cloud and AI-service integrations
  5. For Personal self-hosting, run a W&B server locally with Docker and Python

Limits to check

The Free plan includes five model seats, 5 GB of monthly storage, and 1 GB of monthly Weave data ingestion. Personal is free for one user and personal projects only, with corporate use disallowed. Pro starts at $60/month, includes up to 10 model seats, 100 GB of monthly storage, and 1.5 GB of monthly Weave ingestion, and is intended for teams with fewer than 50 employees.

The Club's review

Weights & Biases is an AI developer platform that combines experiment management, application observability, and governed model and dataset operations. It suits AI teams moving from training and fine-tuning into production, especially when those stages need a shared record. Its broad coverage is compelling, but plan boundaries and deployment caveats matter before committing.

Overview

Founded in 2017 and headquartered in San Francisco, Weights & Biases targets developers building models, agents, and AI applications. W&B Models handles machine-learning experiment tracking and visualization, while Weave covers tracing, evaluation, and monitoring for applications and agents. Registry adds versioned storage for models and datasets, including aliases, lineage, governance, and access controls.

The platform connects with cloud providers and names OpenAI, Meta, Amazon Bedrock, Anthropic, LangChain, and LlamaIndex among its integrations. SaaS, dedicated, and customer-managed deployment options support different operating requirements. The company states certifications to ISO/IEC 27001:2022, ISO/IEC 27017:2015, and ISO/IEC 27018:2019, plus SOC 2 and HIPAA compliance.

Key features

  • Experiment tracking: W&B Models records and visualizes runs, making comparisons across training work practical. Sweeps adds hyperparameter optimization for teams iterating on model performance.
  • AI application observability: Weave supplies tracing, evaluations, and monitoring for AI applications and agents, extending the platform beyond offline model training.
  • Registry and lineage: Registry versions models and datasets and supports aliases, lineage, governance, and access control. That gives teams a controlled handoff from experimentation to shared assets.
  • Lifecycle tooling: Listed capabilities include pipeline orchestration, model serving, managed compute, data and model versioning, deployment tracking, and artifact storage. The combination is more suitable for an end-to-end operating process than for experiment logging alone.
  • Deployment and security: SaaS, dedicated, and customer-managed choices accommodate different control requirements, while the stated certifications and compliance claims will interest regulated organizations.
  • Mobile monitoring: The iOS app is intended for monitoring experiments and training runs. Dedicated and on-premises deployments are not supported in that app, limiting its usefulness for teams using those environments.

Pricing

Weights & Biases uses a freemium model, with a 30-day trial and paid plans starting at $60 per month.

PlanPrice and termsWhat it includes
FreeFree, billed $0/mo5 model seats, 5 GB/month storage, and 1 GB/month Weave data ingestion. It is a useful entry point for small groups, but the storage and ingestion allowances constrain sustained or data-heavy work.
PersonalFree, billed $0/moOne user seat for personal projects only. Corporate use is not allowed, so it is unsuitable for business teams despite the local-server option.
ProStarts at $60/month, billed monthlyUp to 10 model seats, 100 GB/month storage, and 1.5 GB/month Weave ingestion, with priority email and chat support. It is aimed at early-stage teams with fewer than 50 employees; customers exceeding that guideline must move to Enterprise.
EnterpriseCustom plans, invoiced annually upfrontCustomizable seats, storage, and ingestion, enterprise support, and a single-tenant option. This is the route for larger organizations needing scale and a formal support package.
Advanced EnterpriseCustom planSelf-hosted deployment with enterprise features and support, for organizations that cannot use the standard deployment choices.

Platforms

Weights & Biases is offered through its API, web interface, Linux, macOS, Windows, iOS, and self-hosted deployments. The Personal plan permits running a W&B server locally with Docker and Python, but that plan prohibits corporate use.

Who it's for

AI developers and teams handling model training, fine-tuning, application development, and production monitoring are the natural audience. Free fits small experiments with modest data volumes. Pro suits early-stage companies that can stay within its 10-seat and usage allowances. Enterprise and Advanced Enterprise are better for larger or more controlled environments, particularly where customizable quotas, single tenancy, self-hosting, or enterprise support are required.

Pros and cons

Pros

  • One platform spans experiment tracking, AI application monitoring, and model and dataset governance.
  • Registry includes aliases, lineage, access control, and versioning rather than simple artifact storage.
  • Multiple deployment models and stated security certifications support regulated or infrastructure-sensitive teams.
  • A free tier and monthly Pro billing provide a low-commitment starting point.

Cons

  • Free storage and Weave ingestion caps can become restrictive for active projects.
  • Pro is explicitly limited to early-stage teams under 50 employees, forcing a plan transition as organizations grow.
  • Personal cannot be used for corporate work, despite its local self-hosting capability.
  • The iOS app does not support dedicated or on-premises deployments.
  • Enterprise pricing requires annual upfront invoicing, reducing cost transparency and flexibility.

Alternatives

For a broader directory of comparable products, see MLOps Platforms, Model Registry Software, AI LLM Evaluation Tools, Data Version Control Tools, and ML Experiment Tracking Software.

  • MLRun is a free, open-source-oriented option with self-hosted platforms and a managed enterprise offering.
  • Opik suits teams wanting open-source observability and evaluation that can run locally, with a freemium cloud plan.
  • Amazon SageMaker Autopilot is the alternative when pay-as-you-go AWS usage is preferable to a freemium platform plan.
  • Flyte is a free platform option for teams focused on workflow orchestration and self-hosted operation.
  • Outerbounds centers on the open-source Metaflow framework and offers a freemium path.
  • ClearML is worth considering when a 100% open-source self-hosted version is the priority.
  • Databricks Notebooks provides a browser-based notebook environment with a free edition and usage-based paid option.
  • Paperspace Gradient is an alternative for teams seeking notebook-oriented access with a free tier and a listed $8/month Pro user plan.

Verdict

Pick Weights & Biases when your AI team wants a connected system for experiments, application observability, and governed model and dataset assets. Its strongest case is breadth across the machine-learning lifecycle, backed by flexible deployment choices and a usable free starting point. Look elsewhere if your organization is beyond the Pro size guideline without an Enterprise budget, needs corporate self-hosting on a free plan, or depends on mobile monitoring for dedicated or on-premises deployments.

The verdict

Pick Weights & Biases if you need one platform for experiment tracking, AI application monitoring, and governed model and dataset management. Look elsewhere if your team exceeds the Pro size guideline without moving to Enterprise, or if the listed deployment limitations conflict with your requirements.

Questions people ask

Is there a free plan?

Yes. Free costs nothing and includes five model seats, 5 GB of monthly storage, and 1 GB of monthly Weave data ingestion.

How much does Pro cost?

Pro starts at $60/month, billed monthly. It includes up to 10 model seats, 100 GB of monthly storage, and 1.5 GB of monthly Weave ingestion.

Can it be self-hosted?

Advanced Enterprise offers a self-hosted deployment. Personal users can also run a W&B server locally with Docker and Python, but corporate use is not allowed on that plan.

What platforms are supported?

The listing names API, iOS, Linux, macOS, self-hosted, web, and Windows.

Does it provide a model registry?

Yes. Registry stores and versions models and datasets and supports aliases, lineage tracking, governance, and access control.

What the maker publishes

API
W&B provides a full Data API for exporting data logged to the service. source
Artifacts
W&B Artifacts versions datasets and models with lineage and deduplication for reproducible pipelines. source
Audience
The maker presents the platform for AI developers and highlights use cases including model training, fine-tuning, AI application development, and production monitoring. source
Availability commitment
The enterprise SLA commits to at least 99.5% monthly uptime for covered services, excluding AI Features and listed exclusions. source
Company headquarters
LinkedIn lists headquarters at 400 Alabama Street, San Francisco, California 94110, United States. source
Company size
LinkedIn lists Weights & Biases as a privately held software-development company with 201–500 employees. source
Deployment
W&B offers SaaS, dedicated, and customer-managed deployment options. source
Deployment
W&B offers multi-tenant SaaS, isolated dedicated cloud and customer-managed cloud or on-premises deployments. source
Documentation
W&B documentation includes product guides, API references, quickstarts, cookbooks and support articles; the current support site lists 217 W&B Models articles and 19 W&B Weave articles. source
Experiment tracking
W&B Models tracks and visualizes machine learning experiments and supports hyperparameter optimization with Sweeps. source
Experiment tracking
The platform automatically records machine-learning pipeline details and provides contextual visual analysis. source
Founded
2017 source
Founded
LinkedIn lists the company as founded in 2017. source
Headquarters
San Francisco, California, United States source
Integrations
The platform works with cloud providers and names OpenAI, Meta, Amazon Bedrock, Anthropic, LangChain, and LlamaIndex among its integrations. source
Integrations
W&B is designed to integrate with customers' ML stacks and tools and offers managed or self-managed deployment choices. source
Licence
The wandb client repository is released under the MIT License, while the hosted W&B service is commercial. source
Limit: cost and storage
G2 reviewers report that premium pricing and storage or bandwidth costs can rise for small teams and large artifacts. source
Limit: iOS app coverage
The official iPhone app supports SaaS, Free, Academic and cloud Enterprise accounts but not dedicated or on-premises deployments. source
Limit: learning curve
G2 reviewers repeatedly report that the large feature set, advanced configuration and integrations can overwhelm beginners. source
Limits
The pricing page states Pro is for early-stage teams with fewer than 50 employees and that customers exceeding that guideline must transition to Enterprise. source
Mobile app
The maker describes an iOS app for monitoring AI experiments and training runs; the listing says dedicated and on-premises deployments are not yet supported. source
Ownership
LinkedIn identifies the current organization as Weights & Biases by CoreWeave. source
Praise: experiment organization
G2 reviewers repeatedly praise run tracking, metric visualization, experiment comparison, artifact versioning and reproducibility. source

Checked against the maker's pages on 15 Oct 2026.