Microsoft Debuts Decision-1 A $0.042 Decision Model Beating GPT-6 Luna Decision in Speed and Accuracy.
Microsoft Launches Microsoft-Decision-1: High-Speed Decision Engine Outperforming GPT-6 Luna Decision Microsoft has officially introduced Microsoft-Decision-1 , a specialized decision-scoring model engineered to compete directly with models like Jev in routing, classification, and verification tasks. Delivering faster latency and higher accuracy than existing dedicated decision architectures, the model achieves an average accuracy benchmark score of 83.5% , surpassing competing models such as OpenAI GPT-6 Luna Decision , which scored 79.4% across broader evaluation sets. Qwen3.5-9B Foundation, Multi-Benchmark Training, Xbox Case Study, and Pricing Open-Weight Base Architecture and Future Multi-Model Roadmap: Qwen3.5-9B Foundation: The initial release is adapted from Alibaba's Qwen3.5-9B open-weight base model utilizing a post-training strategy similar to Cloudflare’s Clef . Core Engineering Ownership: The model was developed by Microsoft’s core engineering division rather tha...