Friday, November 20, 2026
HomeTechnologyFTA Global's Mark-1 AI model defeats OpenAI GPT-6 Astra, Google Gemini in...

FTA Global’s Mark-1 AI model defeats OpenAI GPT-6 Astra, Google Gemini in chess

Mumbai, November 20, 2026: FTA Global has launched its specialist AI model, FTA Mark-1, at Bharat AI Innovation 2026 in Mumbai, announcing that the model outperformed OpenAI’s GPT-6 Astra and Google’s Gemini 4 Argon and Gemini 3.8 Flash in a series of chess challenges.

Built on the Sarvam-30B open model and trained using FTA Global’s Micro-Judgement Tuning™ (MJT) methodology, Mark-1 won all three of its 12-game chess challenges against the leading global AI models, scoring 28 out of 36 points. The company said Mark-1 runs on a single 80 GB GPU, demonstrating the potential for specialist AI models to be developed with modest computing resources.

Chess challenge results detailed

FTA Mark-1 faced three of the top four models on Kaggle’s chess leaderboard as of November 12, 2026, in independently organized matches. Against OpenAI’s GPT-6 Astra, Mark-1 secured 10–2 (9 wins, 2 draws, 1 loss); versus Google’s Gemini 4 Argon, the result was 9–3 (9 wins, 0 draws, 3 losses); and against Google’s Gemini 3.8 Flash, Mark-1 achieved 9–3 (8 wins, 2 draws, 2 losses). Across all matches, Mark-1 recorded 26 wins, 4 draws, and 6 losses.

FTA Global organized and conducted the matches independently. OpenAI, Google, and Kaggle did not participate in, organize, or endorse the challenges. Each 12-game match used six fixed and balanced opening positions, with both sides playing each opening as white and black. No chess clock was used, and Mark-1 played with its own think-ahead code, while competing models used their makers’ default settings. All records were timestamped on the Bitcoin blockchain prior to result announcements, and the 36 games, including move times, have been published for independent review.

Micro-Judgement Tuning™ methodology

FTA Global’s Micro-Judgement Tuning™ methodology, central to Mark-1’s development, builds on research focused on scoring every possible move, training models to evaluate individual options and select the strongest decisions. Rather than creating a new model from scratch, MJT adapts existing open models for specialized judgement in specific tasks, aiming for efficiency in data, computing, and time.

Mark-1 uses Sarvam-30B as its base, with a lightweight Low-Rank Adaptation (LoRA) layer trained on positions from publicly available Lichess games, scored by the Stockfish chess engine. During gameplay, Mark-1 does not use an external chess engine, instead relying on its own judgement model and FTA’s think-ahead code, which is released openly with the model.

Plans for specialist AI applications

FTA Global intends to extend the MJT methodology beyond chess, targeting business applications such as marketing. From January 2027, the company plans to apply insights from Mark-1 to build marketing intelligence systems for clients, focusing on evaluating possibilities and supporting decision-making. The Mark series is positioned as a suite of specialist intelligence AI models designed to augment human expertise in various industries.

Senthil Kumar Hariram, Founder and Managing Director of FTA Global, said, “The next phase of AI will not be defined only by how large a model is, but by how effectively it can reason, evaluate choices and make decisions. With Micro-Judgement Tuning™, we wanted to explore a different approach to AI development, one where an existing open model can be trained to develop deep expertise in a specific task without requiring enormous amounts of data and computing power. Mark-1 is our opening step in demonstrating this possibility.”

Model availability and company background

FTA Mark-1 is now available on Hugging Face, with the adapter released under the Apache 2.0 license, code under GPL-3.0, and the 36 match games under CC BY 4.0. The company is headquartered in India and employs over 80 professionals, with an advisory network spanning sectors including retail, ecommerce, BFSI, life sciences, healthcare, education, aviation, technology, and academic experts from George Washington University in the USA.


Disclaimer: This article is based on a press release provided by the concerned organisation and has been edited for style and clarity with the assistance of AI tools. BusinessFortnight does not verify, endorse, or take responsibility for the claims, figures, or statements made in the original release.

Ravindra Soni
Ravindra Sonihttps://businessfortnight.com
Senior journalist with 38 years of experience in print and digital media. Former Executive Editor. Founder and Editor of BusinessFortnight, covering business, economy, and policy news from Madhya Pradesh and across India.
RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments