Companies

Weights & Biases

wandb.ai

AI developer platform helping teams build, evaluate, manage, and productionize machine-learning models and applications.

HQSan Francisco, California, United States
Employees201-1000
1 active role
Jobs checked 1h ago
AI / MLAI InfrastructureB2B SaaSAcquired

About

Weights & Biases builds an AI developer and MLOps platform for machine-learning practitioners, research institutions, startups, and enterprises to train models, track experiments, evaluate performance, version data and models, and manage production workflows. Its differentiation is an end-to-end, collaborative system of record designed to make model development reproducible, auditable, and scalable from individual researchers to large teams.

Market

Weights & Biases competes in the MLOps and AI-development-platform market, combining experiment tracking with management of models, pipelines, datasets, evaluation, and governance. It positions itself as a collaborative system of record for ML teams and differentiates through broad framework and workflow integrations, audit/compliance capabilities, and deployment options spanning multi-tenant cloud, dedicated cloud, customer-managed cloud, and on-premises environments. Comet is a particularly direct competitor, positioning its own platform around custom deployments, reliability, and lower pricing.

Target Customers

Weights & Biases targets machine-learning practitioners and teams at AI startups, research institutions, and large enterprises across industries. Its strongest fit is organizations of all sizes with multi-person or large-scale ML programs that need experiment tracking, collaboration, model management, governance, and flexible deployment.

At a Glance

Problem

AI and machine-learning teams must run many experiments while keeping track of data, code, hyperparameters, metrics, failures, and model outputs. Without a shared, searchable record, results are difficult to compare, reproduce, and communicate, leaving researchers and engineers to maintain fragmented “lab notebooks” and perform manual work. The economic pain is slower iteration and inefficient use of scarce training compute: Gretel reported that W&B’s logging and evaluation tools helped it move from queuing 5–10 experiments to running 50–100 in each compute block, while describing a 10x increase in experimentation velocity. The killer use case is therefore experiment tracking and evaluation: automatically capturing each run, seeing performance problems during training, and comparing alternatives quickly enough to build and ship better models.

Product / Service

Weights & Biases is an AI developer platform for developing models and shipping LLM applications through experiment tracking, evaluation, and observability. Its SDK logs metrics, hyperparameters, system metrics, and model artifacts; the platform visualizes and compares runs, supports sweeps for hyperparameter optimization, and provides reports and a registry for versioning and reproducibility. Its broader platform also covers application tracing, output evaluation, cost estimates, LLM comparison, monitoring, guardrails, and workflow automation, extending the product from model development into LLMOps.

W&B is delivered as a multi-tenant SaaS cloud, a dedicated cloud, or customer-managed software running in a customer’s cloud or on-premises infrastructure. That flexibility lets teams adopt a managed service for speed or retain control over data residency, network isolation, and compliance. The benefit is a common operational layer that turns experimental history and production signals into shared, reproducible workflows, helping teams iterate, evaluate, and deploy AI more quickly and confidently.

Market

W&B competes in the MLOps and LLMOps market, positioned today as an AI developer platform spanning model development, application evaluation, and observability. Its competitive set includes open-source and commercial experiment-tracking or AI-lifecycle platforms such as MLflow, Neptune, Comet, and ClearML, with cloud-native services such as Google Vertex AI Experiments also serving adjacent needs. W&B differentiates through a developer-first workflow, rich visualization and collaboration, and an expanding bridge from classic model experimentation to generative-AI application tracing and evaluation.

The evidence indicates substantial traction rather than an unlaunched pre-revenue profile: W&B announced a $50 million investment at a $1.25 billion valuation in 2023, reported growth from 100,000 to more than 700,000 users, and said its platform had tracked nearly 300 million hours of customer experiments. In March 2025, CoreWeave described the platform as being used by more than 1 million AI engineers, and CoreWeave completed its acquisition of W&B on May 5, 2025. The cited evidence does not disclose revenue, but it does establish large-scale adoption, enterprise customers, and a completed strategic exit.

Founders & Leadership

Lukas BiewaldFounder
Senior Vice President of AI Initiatives, CoreWeave
Chris Van PeltFounder
Distinguished Engineer
Shawn LewisFounder
Founder and CTO
Adrian SwanbergVP of Engineering

Funding History

2018-05
Series A$5M

Trinity Ventures, Bloomberg Beta

2019-05
Series B$15M

Coatue

2021-02
Series B$45M

Insight Partners

2021-10
Series C$135M

Felicis Ventures, Bond Capital

2023-08
Series C$50M

Nat Friedman, Daniel Gross

Recent News

2026-07-24partnership
Weights & Biases announces integration with NVIDIA Base Command Platform

Weights & Biases announced an integration with NVIDIA Base Command Platform, a hosted AI development hub for enterprise machine-learning teams.

2026-05-21partnership
Cranium AI and Weights & Biases strategic integration

Cranium AI announced an integration with Weights & Biases that enables enterprises to run AI security, governance, and red-team evaluations on models managed in W&B.

2026-04-17partnership
Robots dream in simulation

Weights & Biases and NVIDIA collaborated on two blueprints focused on scalable reinforcement-learning model training.

2026-03-16product
Announcing the Weights & Biases mobile app: Monitor AI training runs anytime, anywhere

W&B launched a mobile app for checking live experiment metrics and receiving alerts when training runs crash.

2026-03-02product
Introducing W&B Serverless SFT

Weights & Biases announced W&B Serverless SFT, alongside updates and educational content for NLP and generative-AI workflows.

2025-10-30
CoreWeave Acquires Marimo

CoreWeave’s announcement noted that its earlier acquisition of Weights & Biases combined compute and MLOps tools to create an end-to-end AI developer platform.

2025-08-12
CoreWeave Reports Strong Second Quarter 2025 Results

CoreWeave reported completion of its Weights & Biases acquisition and highlighted new products, including capacity available through W&B Models.

2025-09-08product
Humanloop is sunsetting. Migrate to Weights & Biases as an alternative.

W&B promoted its platform as an alternative for Humanloop users seeking AI evaluation, monitoring, and observability after Humanloop’s shutdown.

2025-08-21partnership
Weights & Biases and Azure integration

A Microsoft customer story described how Weights & Biases, through Microsoft for Startups, partnered with Azure to integrate W&B into AI development workflows.

Active Roles

1

Business Model

Weights & Biases monetizes a subscription-based SaaS platform through free, team, and enterprise plans. Enterprise customers are invoiced annually, with additional pricing tied to platform usage such as training, inference, and storage.

Products

Weights & Biases AI development platform for experiment tracking, logging, visualization, and ML lifecycle managementW&B Registry for model managementWeave for evaluating and scoring AI/LLM-driven product changesW&B Sweeps for hyperparameter tuning and experiment exploration

Customers

AstraZenecaBayerBMWCanvaMetaMicrosoftNVIDIAPerplexityRBCSalesforceSamsungSiemensToyotaWayveScaled CognitionUpstageReelDataShellIBMPinterestWoven by ToyotaSambaNovaCapturGretelM-KOPAVisualCortexCoherePandoraMARZInverted AIToyota Research InstituteLyft

Tech Stack

Python SDKMajor machine-learning frameworksTraining and inference environmentsWorkflow orchestration toolsLLM and AI-model development workflows

Competitors

Comet
Datadog
DataRobot