GLiNER-2.5-Decide

Fastino Labs released GLiNER2.5-Decide on September 24, 2026, as an open-weight encoder-based decision model designed to run on CPU hardware rather than GPUs. According to Fastino's release materials and Hugging Face model card, the model carries 340 million parameters and ships under the Apache 2.0 license at the path fastino/GLiNER2.5-Decide. Fastino described it on X as built for fast, deterministic classification. Rather than generating open-ended text, GLiNER2.5-Decide accepts a passage plus a set of user-defined questions or rules and returns a structured decision with probability distributions and confidence scores. Fastino's own benchmark reports a p50 latency of 167.3 milliseconds at batch size 1 and 64 tokens on a 48-vCPU Intel Xeon Platinum 8581C, requiring no GPU. Use cases cited include compliance screening and structured classification workflows where generative LLM nondeterminism is a liability.

Fastino Labs announced GLiNER-2.5-Decide on September 24, 2026, a 340-million-parameter open-weight encoder model designed to replace LLM calls with deterministic, auditable decisions. It targets routing, triage, classification, sentiment, and LLM-as-judge tasks where generative models are overkill. Weights are released under Apache 2.0 on Hugging Face as fastino/GLiNER2.5-Decide. Instead of generating labels, users define a schema of typed questions with permitted answers and cardinality rules; a constrained decoder then resolves contradictions and emits probabilities and confidence metadata. The model is CPU-friendly, reporting 167.3 ms p50 latency at 64 tokens on a 48-vCPU Xeon, and achieves a 60.1% average across 17 datasets, leading on 9.