Fastino Releases GLiNER2.5-Decide: A 340M Open-Weight Decision Model That Runs on CPU
340M open-weight decision model targets agent routing and triage—runs on CPU, ships with confidence scores and constraint metadata.
Why it matters
Fastino's GLiNER2.5-Decide addresses a real gap in agent pipelines: lightweight, CPU-deployable structured decision-making with interpretability built in. Practitioners building guardrails, routing logic, or tool-selection systems now have a measurable alternative to larger models or custom classifiers.
The key facts
11 to knowModel size: 340M parameters
Open-weight release
CPU-deployable (no GPU required)
Output: structured answers + probability distributions + confidence scores + constraint-feasibility metadata
Target use: agent routing, triage, tool selection, guardrails
Source: Fastino Labs via MarkTechPost
340M-parameter open-weight model
Outputs: structured answers, probability distributions, confidence scores, constraint-feasibility metadata
Target use cases: agent routing, triage, tool selection, guardrails
Released by Fastino Labs
Published September 25, 2026
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Fastino Labs has released GLiNER2.5-Decide, a 340M-parameter open-weight decision model. It takes text and a schema of typed questions and returns structured answers. Each answer comes with a probability distribution, a confidence score, and constraint-feasibility metadata. It targets the frequent…