Weights and Biases

aid: wandb
url: https://raw.githubusercontent.com/api-evangelist/wandb/refs/heads/main/apis.yml
name: Weights and Biases
kind: company
description: >-
  Weights and Biases (W&B) is an MLOps and AI developer platform covering the
  full lifecycle of model and LLM application development. W&B Models provides
  experiment tracking, hyperparameter sweeps, artifacts, model registry, and
  reports. W&B Weave provides LLM tracing, evaluation, cost tracking, guardrails,
  and prompt/playground tooling for production AI applications. Three CoreWeave-
  powered serverless capabilities sit alongside the core platform: Serverless
  Inference (OpenAI-compatible API for open-source foundation models),
  Serverless RL (post-training with ART/RULER), and Serverless Sandboxes
  (isolated code execution). The platform exposes a Python SDK, a public REST
  API, and a GraphQL API at api.wandb.ai, with CLI tooling and webhook
  integrations. W&B was acquired by CoreWeave in 2025.
image: https://kinlane-productions.s3.amazonaws.com/apis-json/apis-json-logo.jpg
tags:
  - MLOps
  - Experiment Tracking
  - LLM Observability
  - Model Registry
  - AI Platform
  - Evaluation
  - Tracing
created: '2026-05-23'
modified: '2026-05-23'
specificationVersion: '0.19'
apis:
  - aid: wandb:graphql
    name: W&B GraphQL API
    description: >-
      Primary programmatic surface for W&B Models. The Python public API
      (wandb.Api) speaks GraphQL against api.wandb.ai to query and manage runs,
      projects, sweeps, artifacts, registries, reports, automations, slack and
      webhook integrations, and viewer/entity context.
    humanURL: https://docs.wandb.ai/ref/python/public-api/api/
    baseURL: https://api.wandb.ai/graphql
    tags:
      - GraphQL
      - Runs
      - Artifacts
      - Sweeps
      - Model Registry
    properties:
      - type: Documentation
        url: https://docs.wandb.ai/ref/python/public-api/api/
  - aid: wandb:rest
    name: W&B REST API
    description: >-
      REST endpoints exposed by the W&B platform for ingestion, artifact
      upload, file storage, and integration callbacks. Used internally by the
      wandb SDK and CLI, and available to customers for direct integration.
    humanURL: https://docs.wandb.ai/
    baseURL: https://api.wandb.ai
    tags:
      - REST
      - Ingestion
      - Artifacts
    properties:
      - type: Documentation
        url: https://docs.wandb.ai/
  - aid: wandb:weave
    name: W&B Weave (LLM Observability)
    description: >-
      LLM observability and evaluation platform providing tracing, output
      evaluation, cost estimation, prompt playground, guardrails, and a
      Python and TypeScript SDK. Traces and evaluations are persisted to the
      W&B backend and queryable via the Weave SDK.
    humanURL: https://weave-docs.wandb.ai/
    baseURL: https://api.wandb.ai
    tags:
      - LLM
      - Observability
      - Tracing
      - Evaluation
      - Guardrails
    properties:
      - type: Documentation
        url: https://weave-docs.wandb.ai/
      - type: Repository
        url: https://github.com/wandb/weave
  - aid: wandb:python-sdk
    name: W&B Python SDK
    description: >-
      Official Python SDK (wandb) for logging runs, metrics, gradients, media,
      and artifacts; running sweeps; and interacting with the W&B public API.
      Apache-2.0 licensed.
    humanURL: https://docs.wandb.ai/ref/python/
    baseURL: https://github.com/wandb/wandb
    tags:
      - SDK
      - Python
    properties:
      - type: Repository
        url: https://github.com/wandb/wandb
      - type: Package
        url: https://pypi.org/project/wandb/
  - aid: wandb:cli
    name: W&B CLI
    description: >-
      Command-line interface bundled with the wandb Python package for login,
      sweep orchestration, artifact management, and local agent execution.
    humanURL: https://docs.wandb.ai/ref/cli/
    baseURL: https://docs.wandb.ai/ref/cli/
    tags:
      - CLI
      - Sweeps
      - Agents
    properties:
      - type: Documentation
        url: https://docs.wandb.ai/ref/cli/
  - aid: wandb:serverless-inference
    name: W&B Serverless Inference (CoreWeave)
    description: >-
      OpenAI-compatible inference API for hosted open-source foundation models,
      running on CoreWeave GPU infrastructure with native Weave tracing and
      usage tracking.
    humanURL: https://docs.wandb.ai/guides/inference
    baseURL: https://api.inference.wandb.ai/v1
    tags:
      - Inference
      - OpenAI-Compatible
      - LLM
      - Serverless
    properties:
      - type: Documentation
        url: https://docs.wandb.ai/guides/inference
  - aid: wandb:webhook-automations
    name: W&B Webhook Automations
    description: >-
      Outbound webhook integrations driven by W&B automations. Customers
      register endpoints that W&B POSTs to when configured events fire
      (artifact created, alias added, run state changes, registry events).
    humanURL: https://docs.wandb.ai/guides/automations/
    baseURL: customer-configured
    tags:
      - Webhooks
      - Automations
      - Events
    properties:
      - type: Documentation
        url: https://docs.wandb.ai/guides/automations/
common:
  - type: LinkedIn
    url: https://www.linkedin.com/company/wandbai
  - type: Website
    url: https://wandb.ai/
  - type: Documentation
    url: https://docs.wandb.ai/
  - type: GitHub
    url: https://github.com/wandb
  - type: Weave
    url: https://weave-docs.wandb.ai/
  - type: Pricing
    url: https://wandb.ai/site/pricing/
  - type: Parent
    url: https://www.coreweave.com/
  - type: Plans
    url: plans/wandb-plans-pricing.yml
  - type: RateLimits
    url: rate-limits/wandb-rate-limits.yml
  - type: FinOps
    url: finops/wandb-finops.yml
integrations:
  - name: PyTorch
  - name: TensorFlow
  - name: Hugging Face
  - name: Keras
  - name: scikit-learn
  - name: OpenAI
  - name: Anthropic
  - name: LangChain
  - name: LlamaIndex
  - name: Slack
  - name: GitHub
  - name: Kubernetes
maintainers:
  - FN: Kin Lane
    email: [email protected]