FTA Global, an AI-native Search Engineering and technology company, today unveiled FTA Mark-1, its debut specialist AI model, at Bharat AI Innovation 2026 in Mumbai. Built on India’s open Sarvam-30B model and trained with FTA Global’s Micro-Judgement Tuning™ (MJT) methodology, Mark-1 defeated OpenAI’s GPT-6 Astra, Google’s Gemini 4 Argon and Google’s Gemini 3.8 Flash in three separate 12-game chess challenges. FTA Global chose chess as the model's debut application because it provides a structured environment to evaluate decision-making. Every move requires assessing multiple possibilities, anticipating outcomes and identifying the strongest course of action. This makes chess a measurable benchmark for demonstrating MJT's ability to develop specialized judgement capabilities in an existing AI model. In three separate 12-game challenges against three of the top four models on Kaggle’s chess leaderboard as of November 12, 2026, FTA Mark-1 recorded the following results. All matches were organized and conducted independently by FTA Global.…