Nace AI open source Drex Max v1
The story so far
Nace AI has open-sourced its decision-making model Drex Max v1, which can run locally on Apple Silicon and outperforms Drex v2.0. The model includes bf16 weights, Q8_0 GGUF format, and affine Q4 checkpoints, scoring 67.52 on the public decision-making index (0.4) and achieving 97.3 on InfographicVQA, surpassing Drex v2.0's 61.25. The open-sourced weights of Drex Max allow routing, classification, and policy checks to remain on the local machine. The affine Q4 checkpoint is suitable for Macs with 24 GB or more memory, while Q8_0 GGUF requires 48 GB. On a 48 GB Apple M5 Pro, the performance of Q8_0 GGUF is comparable to H100 bf16.
AI-written from the reports · Updated 18 days ago
Coverage timeline
Follow the coverage to see every side of the story.
- MarkTechPostNace AI Open-Sources Drex Max v1: A 27B Decision Model That Runs Locally on Apple Silicon
Nace AI open-sources its most powerful decision model, Drex Max v1, which can run locally on Apple Silicon hardware. The model includes bf16 weights, Q8_0 GGUF format, affine Q4 checkpoints, and llama.cpp/Ollama branches, scoring 67.52 on the public decision index 0.4, surpassing Drex v2.0 (61.25), and achieving a 97.3 skill score on InfographicVQA.