specification: API Commons Rate Limits specificationVersion: '0.1' schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits provider: Jupyter Notebook providerId: jupyter-notebook created: '2026-05-04' # Provenance stamped 2026-08-11: this artifact was written by the API Evangelist # bulk sweep dated 2026-05-04, not harvested from the provider. See roadmap#35. method: generated modified: '2026-05-05' reconciled: true tags: - Data Science - Interactive Computing - Jupyter - Machine Learning - Notebooks - Python - Rate Limiting description: Project Jupyter does not operate a public hosted API and therefore does not impose first-party rate limits. The Jupyter Server REST and WebSocket APIs run on user-owned infrastructure; any rate limiting is configured by the operator (reverse proxy, JupyterHub, or hosting platform such as Binder / mybinder.org). sources: - https://jupyter-server.readthedocs.io/ - https://jupyterhub.readthedocs.io/ - https://mybinder.readthedocs.io/ responseCodes: throttled: 429 serviceUnavailable: 503 limits: - name: Self-hosted - operator-defined scope: deployment metric: varies limit: not imposed by the Jupyter project; configured at the deployment / proxy layer - name: mybinder.org public service (community-operated) scope: deployment metric: varies limit: see https://mybinder.readthedocs.io/ for fair-use limits on the public Binder federation policies: - name: Self-Hosted Throttling description: Operators typically apply rate limiting at the ingress (NGINX, Traefik, cloud load balancer) rather than inside the Jupyter Server itself. - name: JupyterHub Quotas description: JupyterHub provides per-user concurrency and resource quotas via spawner configuration; these are deployment-defined. - name: Backoff Strategy description: Clients should implement exponential backoff with jitter on 429 / 503 responses. maintainers: - FN: Kin Lane email: kin@apievangelist.com