We describe Panino, a planner submitted to the IPC 2026 Numeric Track. Panino is a greedy best-first search planner built on top of jpddlplus and designed for linear numeric planning problems. Its main feature is a partitioned numeric novelty measure that adapts classical novelty to features induced by numeric and propositional subgoals. Literals appearing in action preconditions and goals induce both Boolean achievement features and distance features measuring progress toward their satisfaction. Novelty is then evaluated over conjunctions of size at most two, considering first achievements of propositional features and strict improvements of numeric distance features within heuristic partitions. The partitions are induced by the subgoaling-based (hadd) heuristic, which is also used as a tie-breaking component in search. We present two configurations: Panino, targeting the Agile track with a cost-maximising final tie-breaker to favour rapid solution discovery, and Panino-anytime, targeting the Satisficing track by using a cost-minimising tie-breaker and iterative cost-bounded restarts to improve solution quality. The resulting planner integrates numeric novelty with heuristic guidance in a lightweight single-planner, single-open-list search architecture for numeric planning.
Winner of the Satisficing and Agile tracks (Overall) at the International Planning Competition 2026 - Numeric Tracks. The Overall category places no restriction on the problems: both Simple Numeric and Linear Numeric expressions are allowed.