Kubernetes v1.35 introduces Extended Toleration Operators, allowing for numeric comparisons in scheduling decisions. This enhancement enables platform teams to set explicit thresholds for tolerations, addressing scenarios where production clusters mix on-demand and spot/preemptible nodes. Previously, Kubernetes only supported exact value matches or existence checks for taints and tolerations, limiting flexibility in workload placement. With the new Greater Than (Gt) and Less Than (Lt) operators, users can now define tolerations based on specific failure probabilities or performance metrics. For example, a critical workload can be configured to only run on nodes with a failure probability below 5%, while a fault-tolerant batch job might tolerate nodes with a failure probability up to 20%. This allows for cost optimization by leveraging cheaper spot instances without compromising on reliability for critical applications. The operators also apply to metrics like GPU compute scores, ensuring that AI workloads only run on appropriately powerful hardware. This functionality streamlines decision-making in dynamic environments and enhances operational safety by keeping most workloads away from risky nodes unless explicitly allowed. The introduction of these operators maintains the safety model of taints and tolerations, as they allow nodes to declare their risk levels while giving pods the option to tolerate them based on defined criteria. This shift from categorical to numeric comparisons significantly improves workload distribution and resource utilization in Kubernetes clusters.