Ant International Launches FalconTST 2.0: A New SOTA Benchmark for Predictive AI in Finance

Ant International has launched Falcon Time-Series Transformer (TST) 2.0, a sophisticated AI model designed to optimize foreign exchange risk management and cross-border payment forecasting. For fintech professionals, this represents a significant shift toward using predictive AI to solve high-stakes liquidity challenges, offering a blueprint for how foundational models can drive capital efficiency in volatile global markets.

What was announced

Ant International introduced FalconTST 2.0, its most advanced time-series transformer model to date. Unlike large language models that focus on text, FalconTST is built specifically to process continuous numerical data, such as transaction flows, account balances, and currency positions. The model is designed to help global businesses determine exactly when funds are needed, in what volume, and in which specific currencies.

The model has achieved a Mean Absolute Scaled Error (MASE) score of 0.666, which Ant International identifies as a State-of-the-Art (SOTA) performance on global public evaluation benchmarks. This precision has resulted in a consistent forecasting accuracy of more than 93% for its users. Initially deployed internally to manage hourly and daily cashflow, the 2.0 version is now being utilized by major global banking partners.

Integration is already underway across several tier-one institutions. Barclays has incorporated the model into its BARX NetFX hedging platform, while Citi is combining FalconTST with its Fixed FX Rates solution. Standard Chartered is utilizing the model alongside its SCALE FX system as part of the PathFin.ai programme, an initiative involving the Monetary Authority of Singapore. Beyond finance, Ant International plans to expand the model’s applications to include supply chain demand forecasting for e-commerce and predictive operations for the aviation industry.

"Large language models have shown how AI can understand and generate information. FalconTST is about another capability that businesses increasingly need: understanding how the world changes over time, and anticipating what comes next. For us, the value of AI is not simply achieving a better forecasting score, but turning that predictive intelligence into real decisions - how much liquidity to prepare, how to manage FX exposure, and how to allocate capital more efficiently. FalconTST 2.0 is an important step toward making predictive AI a foundational capability for global businesses, across payments, accounts and broader financial services,"

Jiang-Ming Yang, Chief Innovation Officer, Ant International.

The companies involved

Ant International is a major player in the global digital payment and financial services landscape, operating under the broader umbrella of Ant Group. The company focuses on enabling merchants to engage with consumers globally through digital payment and marketing solutions. Its partner network includes some of the world's largest financial institutions.

Barclays and Citi are two of the most prominent names in global investment banking and FX services. Barclays operates its BARX platform as a multi-asset electronic trading solution, while Citi remains a dominant force in investor services and fixed-income markets. Standard Chartered, headquartered in the UK but with a deep footprint in Asia, Africa, and the Middle East, is a key participant in regional innovation hubs like Singapore. Deutsche Bank, Germany's largest lender, also joins this cohort in leveraging Ant's AI to refine its liquidity management capabilities. The involvement of the Monetary Authority of Singapore highlights the regulatory and developmental interest in applying AI to systemic financial functions like cross-border settlements.

What FF News has reported before

FF News has closely followed the intersection of predictive technology and institutional finance. We recently covered how Citi Secures US$380 Billion Middle Office Mandate from Aegon Asset Management, highlighting the bank's growing influence in middle-office operations. Additionally, our reporting on Financial Services Resilience Leaders to Examine the Next Phase of Operational Resilience underscores the industry's focus on maintaining stability through technological advancement. We have also tracked the rise of predictive tools in other sectors, such as when High Roller Technologies Taps Thomas Scaria as Chief Product Officer to Lead ROLR Prediction Markets, and monitored broader economic shifts in reports like the JD Power 2026 Study: 60% of U.S. Credit Card Customers Now Classified as Financially Unhealthy.

What this means

This move signals that the "AI arms race" in finance is shifting from generative chatbots to specialized, foundational models that handle core treasury functions. By achieving 93% accuracy in FX forecasting, Ant International is putting pressure on traditional treasury management systems that rely on lagging indicators. For banks like Citi and Barclays, adopting FalconTST 2.0 is a defensive and offensive play: it protects against currency volatility while offering clients superior pricing. The next thing to watch is whether this model becomes an industry standard via its GitHub API trial, potentially commoditizing high-end FX forecasting for smaller fintechs and e-commerce platforms.