Nace AI releases technical report on Drex: A 27-billion parameter decision model uses discriminative reasoning to challenge cutting-edge large language models.
Nace AI has released a technical report on Drex, an open-source decision-making model. The author notes that the 27-billion parameter model focuses solely on making choices rather than generating text, targeting the numerous small decisions required by AI agents. Drex Max performs statistically on par with GPT-6 Sol on Decision Bench v4, outputting probabilities for all options in a single forward pass. It costs $0.06 per 1,000 decisions and has a latency of 183 milliseconds, compared to Claude Opus 5.5 at $10.17 and approximately 3.5 seconds.
Continue looking at this topic
OpportunitiesOne person runs a company in Hangzhou, using AI employees to take on development requests, while also discussing AI team building ↗Hotly discussed viewsWith more and more plugins for coding agents, how can teams pick the one that truly saves effort? ↗Related contentWhat is Claude Code? Understanding how the model participates in project development ↗Open-source projectGlen-Web-App/best-agent-memory-mcp ↗Return to original text
A short introduction is retained here; the full text from the source website is not reposted. For specific wording and subsequent updates, please refer to the original article.
Open AIHOT original