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: Langfuse
providerId: langfuse
created: '2026-05-08'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-08, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-08'
reconciled: false
tags:
- AI
- LLM
- Observability
- Open Source
- Evaluations
- Rate Limiting
- Quotas
- Throttling
description: Langfuse cloud throttling is primarily quota-based on monthly units (traces,
  observations, scores). Enterprise accounts can negotiate custom rate limits. Self-hosted
  installations are limited only by the operator's own infrastructure.
notes: Langfuse does not publish per-second request-rate ceilings for cloud. Watch for 429
  responses and backoff. Self-hosted has no platform-imposed limits.
sources:
- https://langfuse.com/pricing
- https://langfuse.com/docs
responseCodes:
  throttled: 429
limits:
- name: Hobby Cloud Monthly Units
  scope: account
  metric: unit
  limit: 50000
  timeFrame: monthly
  notes: Hard cap for the free Hobby tier.
- name: Core/Pro/Enterprise Monthly Units (Included)
  scope: account
  metric: unit
  limit: 100000
  timeFrame: monthly
  notes: Included monthly allowance; overage billed at $8/100k with volume discounts.
- name: Enterprise Custom Rate Limits
  scope: account
  metric: varies
  limit: contract
  notes: Enterprise customers may negotiate custom per-second or per-minute API rate ceilings.
- name: Self-Hosted
  scope: instance
  metric: none
  limit: -1
  notes: No platform-imposed limits when self-hosted; bounded by operator infrastructure.
policies:
- name: Backoff Strategy
  description: Honor 429 responses with exponential backoff and jitter; SDKs handle retry
    automatically.
- name: Trace Batching
  description: SDKs batch trace ingestion calls before flushing, reducing per-request overhead.
- name: OpenTelemetry Path
  description: Use the OTel ingestion endpoint where possible; it is the recommended,
    higher-throughput path.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com