Nace AI Open-Sources Drex 1.5: A 9B Decision Model That Scores Options, Not Text

Nace.AI has released Drex 1.5 as an open-source model, marking a notable departure from the dominant paradigm of generative text models. Rather than producing free-form text, Drex 1.5 assigns a probability score to each candidate option in a single forward pass, positioning it as a purpose-built decision engine rather than a general-purpose language model.

The model carries 9 billion parameters and supports context windows of up to 128K tokens, which places it in a practical range for many enterprise and agent workflows. Its reported score of 58.08 on Decision Index 0.3.1 provides a benchmark reference, though the metric itself is relatively new and may not yet be widely adopted across the community.

The decision-scoring approach could be relevant for applications where structured choice selection matters more than open-ended generation, such as recommendation systems, routing logic, or multi-option evaluation pipelines. By returning calibrated probabilities directly, the model may reduce the need for post-processing or parsing of generated text.

Open-sourcing the model allows researchers and developers to inspect, fine-tune, and integrate Drex 1.5 into their own stacks. Whether the decision-model paradigm gains traction will depend on how well it performs in real-world tasks compared to conventional generative approaches and whether the broader ecosystem develops supporting tooling and benchmarks.

This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.