BlockBeats News, July 9th, the DMind Benchmark launched by the Minara team in 2025 was recently officially accepted by the Datasets & Benchmarks Track of KDD 2026, becoming the first cross-disciplinary large-scale model evaluation benchmark in the intersection of digital assets and AI to undergo peer review and enter a top international conference (not a workshop or demo). KDD is a top international conference in the field of data mining and machine learning, with an acceptance rate of about 29% for this track at this year's conference.

The DMind Benchmark measures the model's multi-step reasoning ability in real-world digital asset scenarios, consisting of 3,154 objective questions and 389 open-ended tasks that have been expert-reviewed, covering 9 major subfields including DeFi, Tokenomics, and smart contract security. It has systematically evaluated 31 cutting-edge large models such as GPT, Claude, and Gemini. The dataset once ranked #1 on the Hugging Face platform's Trending list, with over 13,000 cumulative downloads.

A reevaluation in July 2026 showed that the progress of the six latest flagship models in this benchmark did not bridge the deep gaps in the field's capabilities. Leveraging this benchmark and DMind, which is building a dedicated intelligent foundational layer for the digital asset domain, is enabling AI to truly understand the world of digital assets.