{
  "type": "article",
  "title": "Mostik's New Mathematical Bridge Empowers Small AI Models to Achieve High Performance Without Text Generation",
  "summary": "A novel mathematical approach created by startup Mostik allows artificial intelligence models to exchange internal weights directly without producing text, enabling small models to run efficiently with near-frontier capability.",
  "content": "A team of mathematicians working at an artificial intelligence startup has developed a novel method for AI models to communicate directly with one another. The startup, named Mostik after the Russian word for bridge, designed an approach that links distinct neural networks by using the raw mathematical values embedded within their internal weights. In machine learning, weights are the underlying numerical parameters that dictate how an input prompt is transformed into a generated response. By establishing a direct connection between these values, the capability of a massive frontier model can be shared with a smaller companion model, upgrading its performance rapidly and at a dramatically reduced computational cost.\n\nARC-AGI 3 Success and Benchmark Performance\nUsing this weight-bridging technique, Mostik developed an AI system that reached the top of the ARC-AGI 3 benchmark leaderboard, an evaluation widely regarded as one of the toughest tests for machine intelligence. While the startup has kept the full architectural specifics confidential due to the ongoing contest, they built a public demonstration linking two open-weight models developed in China. The team connected the largest iteration of GLM-5.2, which contains 753 billion parameters, with a 4-billion-parameter version of Qwen-3.5 designed to execute locally on consumer smartphones. The resulting hybrid system achieved performance exactly midway between the small and large models, while operating at merely one-twentieth the computational cost of the full GLM model.\n\nBypassing Text Generation Bottlenecks\nIt is widely recognized across machine learning that combining outputs from multiple models, a practice known as ensembling, produces superior results compared to running a single model in isolation. Mostik Chief Executive Officer Sasha Malysheva compared the phenomenon to a classic mathematical demonstration regarding collective estimation. In statistics, combining and averaging independent guesses from a crowd of non-experts often yields a more accurate measurement of an object's weight than relying on a single expert opinion. Combining output from several AI models functions similarly. However, traditional ensemble techniques rely on generating text from one model and feeding it as a prompt into another, an operation that consumes significant time, memory, and energy. Mostik eliminated this text generation bottleneck by enabling silent weight-level mathematical communication. If widely adopted, this strategy could substantially increase the utility of open-weight systems competing against proprietary models from companies like Anthropic and OpenAI.\n\nShift Toward Specialized Domain Architectures\nSasha Malysheva believes that coupling diverse, smaller models will prove to be a more viable path forward for artificial intelligence than relying strictly on raw scaling. Rather than building massive monolithic models trained on vast internet datasets, future improvements may rely on combining general-purpose networks with specialized systems tailored for fields such as biology or physics. Vladimir Arustamian, tech lead at the AI software company Lovable, noted that Mostik's team managed to deliver a working weight-bridging pipeline in a matter of months, delivering progress that observers expected would take years of research.\n\nEfficiency Gains and Brain Mechanism Comparisons\nKarl Tuyls, a former computer scientist at Google DeepMind, highlighted that Mostik's framework allows developers to approximate the capabilities of giant models without running the entire inference loop through them. By running a compact model alongside a bridged network, operations achieve major performance uplifts while staying resource-efficient. Meanwhile, Stanislav Smirnov, a professor at the University of Geneva and recipient of the 2010 Fields Medal who serves as Mostik's chief scientist, emphasized the theoretical significance of the work. Smirnov noted that finding a common mathematical framework between disparate neural networks is extremely challenging because formal mathematical languages for model interaction do not yet exist. Bridging the gap empirically provides a practical solution and could offer new insights into how artificial models function compared to human cognition. Deeper mathematical study may eventually reveal shared patterns in how artificial neural networks and the human brain process complex reasoning tasks.\n\nOvercoming Doubts and Academic Skepticism\nSasha Malysheva's interest in mathematics began in her youth after her older brother suggested she lacked the skill to solve Math Olympiad problems. Motivated to prove him wrong, she mastered advanced mathematics and secured admission to an elite educational institution in St. Petersburg. When she later proposed the concept of creating a direct mathematical bridge between AI model weights, several industry peers expressed skepticism, suggesting the task would be too complex to execute. By turning that skepticism into drive, Malysheva and her team successfully demonstrated that direct weight communication can make frontier-grade intelligence far more accessible and cost-effective across the industry.\n\nWhat this means for you\nThis technology could fundamentally alter how AI models are deployed, lowering operational expenses while raising capabilities.\n\n• For AI Developers Across India: Indian developers and startups can now build high-performing AI tools on modest budgets. Powerful AI services can be delivered without spending heavily on massive cloud computing hardware.\n• For Smartphone and Mobile Users: As models like Qwen-3.5 gain higher intelligence locally, mobile applications will operate faster and with greater accuracy. Users can access advanced AI features directly on their devices without needing constant cloud connectivity.\n• For Enterprise and Researchers: Organizations will no longer need complete reliance on expensive proprietary APIs, making it far cheaper to build specialized models for fields like biology and physics.\n• For Cost Efficiency: Operational costs for running near-frontier AI intelligence can drop to one-twentieth of full model costs, enabling smaller businesses to integrate advanced AI workflows.\n\nQuestions & Answers\n\n1. How does Mostik's new technology function?\nThe technology links distinct AI models directly using the mathematical values in their internal weights, bypassing intermediate text output completely.\n\n2. Which AI models were paired in the demonstration experiment?\nThe team created a mathematical bridge between GLM-5.2 with 753 billion parameters and a mobile-capable 4-billion-parameter version of Qwen-3.5.\n\n3. What were the cost and performance results of the hybrid system?\nThe hybrid system operated at one-twentieth the cost of the full GLM model, achieving performance exactly midway between the two models.\n\n4. Who serves as the chief scientist at Mostik?\nStanislav Smirnov, a professor at the University of Geneva and recipient of the 2010 Fields Medal, is Mostik's chief scientist.\n\nInspiration & Lessons\nSasha Malysheva's story offers valuable insights on persistence, intellectual curiosity, and defying skepticism.\n\n• Turning Doubt into Motivation: When her older brother suggested she could not solve Olympiad math problems, Sasha used the challenge as a catalyst to master mathematics.\n• Overcoming Skepticism: Facing criticism from peers who claimed the complex bridge approach was too difficult for a young female mathematician, she chose to demonstrate success through tangible results.\n• Solving Core Efficiency Problems: Rather than relying on brute-force computing budgets, Malysheva focused on mathematical ingenuity to achieve breakthrough AI performance.",
  "url": "https://trendkia.com/en/ai/mostik-ke-nae-ganitiya-brija-se-chhote-ai-modala-paenge-bari-kshamata-bina-teksta-janareshana-ke-apasa-men-jurenge-ai-26673",
  "category": "AI",
  "publishedAt": "2026-09-02",
  "tags": [
    "AI Models",
    "Mostik",
    "Machine Learning",
    "Artificial Intelligence",
    "Tech News",
    "AI Research"
  ],
  "language": "en",
  "site": "TrendKia"
}