Shivik Labs, a deep-tech research and engineering firm focused on foundational artificial intelligence, has announced the release of its latest research paper introducing TRIDENT, a new reasoning framework aimed at enabling autonomous self-improvement in large language models. The framework seeks to overcome what researchers describe as the “static intelligence” limitation of current AI systems.

According to the company, most large language models today rely on scaling data, parameters, or fine-tuning cycles to improve performance, while their reasoning processes remain largely unchanged. TRIDENT introduces a different approach by treating reasoning as a structured search problem rather than a single linear sequence of outputs. This allows models to explore, evaluate, and refine multiple reasoning paths before arriving at a solution.

Shivik Labs reported that it has open-sourced the TRIDENT framework along with a model built on Qwen3-4B. Using this setup, the model’s performance on the GPQA benchmark improved from 28.28 percent to 42.42 percent, marking a gain of 14.14 percentage points without additional fine-tuning or new training data. Similar improvements were observed across benchmarks such as GSM8K, MMLU, Winogrande, and ARC-C, indicating stronger reasoning and problem-solving capabilities.

The TRIDENT architecture combines three key elements: a tree-based reasoning structure that explores multiple logical paths simultaneously, a graph neural network that evaluates and scores intermediate reasoning states, and a self-generative learning loop that allows the model to audit and refine its own reasoning without human-authored explanations. Together, these components enable models to identify errors, discard weak reasoning paths, and reinforce successful strategies autonomously.

Shivik Labs said the framework is already being tested within its construction execution and control platform to validate real-world applicability. Founder Abhisek Khandelwal stated that India’s scale and operational complexity provide an ideal environment to build resilient reasoning systems capable of handling real-world challenges.

The company plans to develop an indigenous reasoning-focused AI model, with a 2B-parameter version targeted for release in early 2026. Shivik Labs has also launched collaborative pilot programs for organizations dealing with reasoning-intensive problems, while making its research and model artifacts publicly available to encourage wider adoption and innovation.