venice_ai.core.config.enterprise
Enterprise configuration models for Venice AI.
These models are optional and only needed for advanced deployments: circuit breaker, state management, scheduler, and metrics.
CircuitBreakerConfig Objects
class CircuitBreakerConfig(BaseModel)
Configuration for circuit breaker pattern implementation.
Enterprise feature: Wraps backend/account calls with a circuit breaker that
opens after failure_threshold failures, preventing cascading failures.
CircuitBreakerConfig.validate_reset_timeout
@field_validator("reset_timeout")
@classmethod
def validate_reset_timeout(cls, v: float) -> float
Ensure reset timeout is reasonable.
StateConfig Objects
class StateConfig(BaseModel)
Configuration for the unified state management system.
Enterprise feature: Persists rate-limit state, reservations, and account health for the adaptive scheduler. Not required for basic SDK usage.
Key areas: cache policy (write-through/back/around), TTL settings, batch sizes, concurrency control, and cleanup intervals.
SchedulerConfig Objects
class SchedulerConfig(BaseModel)
Unified configuration for the Venice AI request scheduler.
Enterprise feature: Only active when RateLimiterMode.ADAPTIVE is selected
(VeniceAIConfig.rate_limiter.mode = RateLimiterMode.ADAPTIVE).
Requires: pip install venice-ai[adaptive] and a Redis backend.
For basic SDK usage the default RateLimiterMode.SIMPLE mode does not use
this scheduler at all — only HttpClientConfig is needed.
SchedulerConfig.validate_strategy
@field_validator("strategy")
@classmethod
def validate_strategy(cls, v: str) -> str
Validate scheduling strategy.
SchedulerConfig.validate_overflow_policy
@field_validator("overflow_policy")
@classmethod
def validate_overflow_policy(cls, v: str) -> str
Validate overflow policy.
MetricsConfig Objects
class MetricsConfig(BaseModel)
Configuration for metrics collection and cleanup.
Enterprise feature: Controls the internal Prometheus-compatible metrics store used by the observability subsystem. Only needed when metrics export is enabled.