{
  "id": 5143982,
  "title": "These Russian Mathematicians Taught AI Models How to Talk to Each Other Without Using Words",
  "url": "https://urgent.news/2026/09/02/these-russian-mathematicians-taught-ai-models-how-to-talk-to-each-5143982",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-02T18:20:00.000Z",
  "source": {
    "name": "Wired",
    "slug": "wired",
    "url": "https://www.wired.com/story/russian-startup-mostik-ai-models-communication/"
  },
  "original_language": "en",
  "account": "Three Russian mathematicians at a startup called Mostik, whose name means \"bridge\" in Russian, have developed a novel approach to enable different AI models to communicate with each other without using words. Their technique leverages the mathematical values present in the weights of AI models, which determine how a prompt gets converted into an output. By using this method, the performance of a larger model can be transferred to a smaller model, significantly boosting its intelligence with greater efficiency.\n\nMostik's work has resulted in a hybrid system that combines a 753 billion-parameter GLM-5.2 version with a 4 billion-parameter Qwen-3.5 model that can run on a mobile device. The resulting system costs one-twentieth of the full GLM model and performs exactly halfway between the two models. Mostik's CEO, Sasha Malysheva, explained that combining the outputs of several AI models often yields better results, similar to how guessing the weight of a pig becomes more accurate when multiple people participate in the guessing process.\n\nThe startup's innovative approach has the potential to increase the value of open-weight models, allowing them to better compete with the closed, proprietary models from leading AI labs such as Anthropic and OpenAI. Mostik's technique of enabling AI models to communicate without generating text output could also lead to the development of more specialized models for specific domains like biology and physics, as suggested by tech lead Vladimir Arustamian.\n\nThe team's progress has been impressive, as they have already achieved results that could have been considered years away. Mostik's approach has been described as a \"no-brainer\" for those looking to maximize the efficiency of their model operations. The company's chief scientist, Stanislav Smirnov, a 2010 Fields Medalist, acknowledged that finding a common mathematical language between two AI models is challenging. However, Mostik's work could potentially reveal new insights into how AI models function and compare to human reasoning in solving complex problems.",
  "summary": "A startup called Mostik has a wild new approach to combining the capabilities of AI models.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "Wired Business",
        "title": "These Russian Mathematicians Taught AI Models How to Talk to Each Other Without Using Words",
        "url": "https://urgent.news/2026/09/02/these-russian-mathematicians-taught-ai-models-how-to-talk-to-each",
        "published": "2026-09-02T18:20:00.000Z"
      }
    ]
  },
  "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."
}