Agno · Rate Limits

Agno Agi Rate Limits

Agno does not publish fixed numeric rate limits for the AgentOS API, because AgentOS is software you run yourself rather than a shared multi-tenant service Agno operates. Throughput on a self-hosted instance is bounded only by your own infrastructure (CPU, memory, concurrency of the ASGI server) and by the rate limits of whichever LLM provider(s) you configure the agents to call - those provider limits are documented by the provider, not by Agno. The hosted os.agno.com Control Plane gates concurrent "live connections" and seats by pricing plan rather than by a request-per-minute cap.

Agno Agi Rate Limits is the machine-readable rate-limit profile for Agno on the APIs.io network, conforming to the API Commons Rate Limits specification.

It captures 4 rate-limit definitions, measuring requests, connections, seats, and tokens/requests.

The profile also includes 2 backoff/retry policies defined and response codes documented for throttled.

Tagged areas include AI, Agents, Multi-Agent, Rate Limiting, and Quotas.

4 Limits Throttle: 429
AIAgentsMulti-AgentRate LimitingQuotas

Limits

AgentOS API Requests deployment
requests
not published (self-hosted; infrastructure-bound)
No fixed numeric request-rate limit is documented; bounded by your own deployment's compute and concurrency.
Control Plane Live Connections account
connections
per plan (1 included on Pro, additional at $95/month each)
Gates concurrent AgentOS instances connected to the hosted os.agno.com Control Plane, not per-request throughput.
Control Plane Seats account
seats
per plan (4 included on Pro, additional at $30/month each)
Governs number of users who can access the hosted Control Plane, not API throughput.
Underlying Model Provider Limits model provider
tokens/requests
set by the configured LLM provider
Agent, team, and workflow runs are ultimately bounded by the rate limits of whichever model provider (OpenAI, Anthropic, etc.) is configured for that agent.

Policies

Bring Your Own Model Limits
Because AgentOS is self-hosted and model-agnostic, effective throughput is governed by the rate limits of the LLM providers you configure, not by Agno.
Backoff Strategy
Clients should implement exponential backoff with jitter and honor Retry-After when a configured model provider or a fronting proxy returns 429.

Sources