Ant International launches FalconTST 2.0 for banks

Ant International has launched Falcon Time-Series Transformer Model 2.0, which Barclays, Citi, Deutsche Bank and Standard Chartered have adopted for foreign exchange and liquidity forecasting.

The model is designed for time-series forecasting in finance, helping institutions predict cash flows, liquidity needs and currency exposure from changing numerical data. FalconTST 2.0 is Ant International's most advanced version so far and is used in cross-border payments and foreign exchange risk management.

According to Ant International, the latest version recorded a Mean Absolute Scaled Error score of 0.666 on a global public benchmark for time-series foundation models. That result placed FalconTST 2.0 at the top of the leaderboard on that measure, ahead of rival models from large technology groups.

The launch reflects a broader push to apply artificial intelligence beyond text and image generation to prediction-based work. In payments and treasury operations, the value lies in estimating when funds will be needed, how much will be required and in which currencies, because those decisions affect hedging, liquidity and capital allocation.

Ant International first deployed FalconTST internally to manage cash flow and foreign exchange exposure on hourly, daily and weekly cycles. It later integrated the model with banking partners, which now use version 2.0 in their own systems linked to work for Ant International and its clients.

Barclays has integrated the model into its BARX NetFX platform, while Citi combines it with its Fixed FX Rates offering. Standard Chartered uses the model alongside its SCALE FX system as part of the PathFin.ai programme run by the Monetary Authority of Singapore.

Ant International also named Deutsche Bank as one of the banks that have integrated the model for cash flow forecasting and foreign exchange management. It said current deployments have consistently delivered forecast accuracy above 93%.

Finance focus

Time-series transformer models are built to analyse sequences of numerical data over time. That makes them relevant to financial institutions, airlines, retailers and logistics groups that need to forecast patterns such as payment flows, sales demand, booking revenue and operational costs.

Ant International said traditional forecasting systems are often built for a single task, such as retail sales or airline demand. By contrast, FalconTST is trained to recognise broader patterns, including cycles, trends, seasonality and abrupt shifts across several industries, allowing the same model structure to be used in different settings.

Ant International also highlighted technical changes in version 2.0 aimed at handling missing data, learning patterns across industries and processing multiple time frequencies in one architecture. In practical terms, that means distinguishing between an actual zero and missing data, and working with everything from second-by-second payment information to monthly economic indicators.

Wider use

Beyond banking, Ant International said the model is being used in aviation for liquidity and foreign exchange management and is expanding into eCommerce and logistics. Those sectors often collect revenue and incur costs in different currencies, creating exposure to exchange-rate swings if forecasts are inaccurate.

For airlines in particular, predicting foreign currency inflows and outflows matters because overestimating receipts can lead to excessive hedging, while underestimating them can leave a business exposed. The same logic applies to eCommerce platforms handling cross-border transactions at scale.

Jiang-Ming Yang outlined the company's view of the market for predictive artificial intelligence.

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

Kelvin Li,  General Manager of Platform Tech and Senior Vice President, Ant International, said the newer version had broadened the product's reach beyond Ant International's own operations.

"FalconTST helps global businesses - including our own - manage complex cash flow and FX exposure, so they can manage cross-border transactions with greater confidence. With FalconTST 1.0, clients saw real operational value and cost savings from better forecasting. With FalconTST 2.0, enhanced accuracy and precision let us extend those benefits to our banking partners as well as a broader range of customers across fast-moving sectors like eCommerce, travel and fintech," said Li.