{
  "id": 20483068,
  "title": "Microsoft’s new coding model promises near-Opus performance at 5% of the cost",
  "url": "https://urgent.news/2026/11/17/microsofts-new-coding-model-promises-near-opus-performance-at-5-of",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-17T19:14:48.000Z",
  "source": {
    "name": "The New Stack",
    "slug": "the-new-stack",
    "url": "https://thenewstack.io/microsoft-mai-code-2/"
  },
  "original_language": "en",
  "account": "On Tuesday at the Ignite conference in San Francisco, Microsoft unveiled two new coding models, Code 2 and Code 2 Flash, which promise to match the performance of models like GPT-6.1 Sol and Claude Opus 5.5 at significantly lower costs. The company emphasizes that its MAI Code 2 model is at the cutting edge of price-performance for enterprise coding tasks. According to Microsoft, MAI Code 2 achieves 98% of the performance of GPT-6.1 Sol at 40% of the cost and 96% of Claude Opus 5.5 at just 5% of the cost. The models have also been fine-tuned to create effective dashboards and visualizations, with a particular focus on visual coding. In contrast, GPT-6.2 Astra and Sol, released by OpenAI on the same day, show improved performance in certain benchmarks, though Anthropic's models are still known for their high costs. Microsoft's MAI Code 2 Flash is said to match Anthropic's Claude Haiku 5.5 and GPT-6 Luna on certain benchmarks at a fraction of the cost. The company argues that these models are specifically tuned for everyday coding tasks rather than specialty areas like fundamental science research or agentic work. Microsoft's approach contrasts with OpenAI's shift from building specialized models to creating more general ones. Instead, Microsoft aims to offer efficient and cost-effective specialized models, such as MAI Work Flash, which is designed to work with Microsoft 365 and Fabric IQ products. These models are now accessible via GitHub Copilot and Microsoft Foundry, with additional support coming soon to platforms like OpenRouter, Vercel, Hermes, Open Code, and Kilo.",
  "summary": "At its Ignite conference in San Francisco on Tuesday, Microsoft announced a number of new AI models, including two new The post Microsoft’s new coding model promises near-Opus performance at 5% of the cost appeared first on The New Stack .",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "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."
}