Sakhanda Wire
NVDA $154.51 -1.04% MSFT $506.49 -2.22% GOOGL $281.27 -0.86% META $621.51 -0.78% AMZN $256.90 -0.33%
← Back to the news

Fastino releases 340M open-weight model for structured decisions

Fastino Labs has released GLiNER2.5-Decide, a 340M-parameter open-weight model for operational decisions in AI agent pipelines. It accepts text and a schema of typed questions, then returns structured answers with probability distributions, confidence scores and constraint-feasibility metadata. The weights are available under Apache 2.0, can run on CPU or GPU, and are suitable for air-gapped environments. Hosted inference and fine-tuning are also available through the GLiNER API.

The model is a non-generative classifier based on a DeBERTa-v3-large encoder. A schema can define permitted answers, single- or multi-label outputs, ordered values, examples, descriptions and rules connecting questions. Its constrained decoder selects the highest-scoring assignment allowed by those rules, including implications, exclusions and cardinality limits. Fastino says the model is intended for routing, triage, tool selection and guardrails, rather than reasoning or answering open-ended questions.

On Fast Decisions, an internally generated held-out suite of 5,100 examples across 17 datasets, GLiNER2.5-Decide achieved 60.1% average exact-match accuracy and led 9 datasets. It scored 75.3% on support intent and 64.3% on banking intent. In batch-one testing with short inputs, reported p50 latency was 167.3 ms on a 48-vCPU Intel Xeon Platinum 8581C and 38.3 ms on an NVIDIA V100.

Written by Sakhanda Wire AI from reporting by MarkTechPost. Facts, quotes and full context are in the original article below.

Originally published by MarkTechPost on

Read the original on MarkTechPost ↗

The reporting and the image belong to MarkTechPost and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news