Nace.AI has open-sourced Drex 1.5, a 9B-parameter decision model that takes a different approach from conventional text-generating LLMs. Instead of producing free-form text, it assigns a probability to every option in a single forward pass, making it purpose-built for decision-making tasks.
According to the release, Drex 1.5 scores 58.08 on Decision Index 0.3.1 and supports a context length of up to 128K tokens. That combination of long-context reading and direct option scoring could make it useful for applications that need fast, structured choices from large inputs.
The model is now available as open source, allowing developers to integrate or fine-tune it for their own decision workflows. The single-pass scoring design may also reduce compute overhead compared to generating and parsing text for each candidate option.