{
  "id": 13351669,
  "title": "Microsoft leans on open weight model from Chinese AI lab to challenge Jev",
  "url": "https://urgent.news/2026/10/10/microsoft-leans-on-open-weight-model-from-chinese-ai-lab-to-challenge",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-10T07:10:00.000Z",
  "source": {
    "name": "The Register Science",
    "slug": "the-register-science",
    "url": "https://www.theregister.com/ai-and-ml/2026/10/10/microsoft-leans-on-open-weight-model-from-chinese-ai-lab-to-challenge-jev/5302473"
  },
  "original_language": "en",
  "account": "Microsoft has entered the competitive realm of decision models, joining a growing list of companies that have developed their own models to offer unique decision-making capabilities. These models have become increasingly popular due to their ability to provide structured outputs for software to act upon, making them a valuable tool for various business applications.\n\nThe buzz around decision models gained momentum with the announcement of Jev, a large language model developed by TypeSafe AI. Jev's speed, affordability, and response constraints have made it an attractive option for businesses seeking reliable decision-making models. Other companies, such as OpenAI, Cloudflare, Strands, Liquid AI, Perplexity, Snowflake, Surogate Rune, and H2O.ai, have also joined the race by introducing their own decision models.\n\nMicrosoft has now joined the competition with its own decision model, Microsoft-Decision-1. This model, offered via Microsoft Foundry and soon via OpenRouter, is built on the Qwen3.5-9B model developed by Alibaba Cloud, a Chinese tech giant. Microsoft claims that its decision model outperforms competitors in several areas, including speed, accuracy, and cost-effectiveness.\n\nMicrosoft-Decision-1 is 2.5 times faster than H2O-Lightning-4B and 2.8 times faster than Jev, making it a competitive choice for businesses looking for efficient decision-making models. The model also boasts an impressive accuracy rate of 83.5 percent on 36 benchmarks and a confidence score of 92.2 percent, placing it second only to Quyet-1.0-Large in this metric. Additionally, Microsoft-Decision-1 is significantly more affordable than OpenAI's GPT-6 Sol, costing just $0.042 per million input tokens and offering free output tokens.\n\nAs agentic AI continues to evolve, the importance of selecting the right model for specific tasks becomes increasingly crucial. Microsoft's entry into the decision model market demonstrates the growing demand for specialized models that can deliver structured outputs and high performance, ultimately unlocking a wide range of useful tasks for businesses.",
  "summary": "The first version of Microsoft-Decision-1 is based on Qwen3.5-9B, but the next will sport homegrown tech, Redmond reassures",
  "key_points": [
    "Microsoft enters decision model market with Microsoft-Decision-1",
    "Model built on Qwen3.5-9B from Alibaba Cloud, a Chinese AI lab",
    "Outperforms Jev in speed, accuracy, and cost-effectiveness"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "The Register",
        "title": "Microsoft leans on open weight model from Chinese AI lab to challenge Jev",
        "url": "https://urgent.news/2026/10/10/microsoft-leans-on-open-weight-model-from-chinese-ai-lab-to-challenge-13353859",
        "published": "2026-10-10T07:10:00.000Z"
      }
    ]
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}