Ant International has launched FalconTST 2.0, a time-series AI model built for foreign exchange risk management and cross-border payment forecasting. The model, which achieves a Mean Absolute Scaled Error score of 0.666 on a leading public benchmark, already powers cashflow and FX decisions inside major banks including Barclays, Citi, and Standard Chartered. For finance professionals, the release signals a shift: predictive AI is moving from experimental chatbot use cases to core treasury functions where accuracy directly affects capital efficiency.
What FalconTST 2.0 does differently
Unlike large language models that parse text, time-series transformers are built for continuous numerical data - transaction flows, account balances, settlement volumes, and currency positions. FalconTST processes that data to help businesses determine when funds are needed, in what volume, and in which currencies. That capability matters for any organization managing cross-border payments, where liquidity shortfalls and FX exposure can be costly.
Ant International reports the model has delivered consistent forecasting accuracy of more than 93% for its users. It was initially deployed internally at Ant International to manage hourly, daily, and weekly cashflow and FX exposure. Now, with version 2.0, it is integrated into the hedging and pricing platforms of tier-one financial institutions: Barclays runs it on the BARX NetFX hedging platform, Citi pairs it with its Fixed FX Rates solution, and Standard Chartered uses it alongside its SCALE FX system as part of the PathFin.ai programme, an initiative with the Monetary Authority of Singapore. Deutsche Bank is also adopting the model for liquidity management.
What the adoption pattern reveals
The involvement of multiple global banks points to growing demand for specialized forecasting models in financial services. Barclays, Citi, and Standard Chartered are using FalconTST for FX risk management for e-commerce platforms and airlines, according to Ant International. Beyond finance, the company plans to expand the model's applications to supply chain demand forecasting and predictive operations for aviation.
Jiang-Ming Yang, Chief Innovation Officer at Ant International, framed the value in operational terms: "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."
Kelvin Li, General Manager of Platform Tech and Senior Vice President at Ant International, said the earlier version of the model delivered "real operational value and cost savings from better forecasting," and that version 2.0 extends those benefits to banking partners and customers in sectors like e-commerce, travel, and fintech.
What this means for finance teams
The development pressures traditional treasury management systems that rely on lagging indicators. FalconTST's 93% forecasting accuracy - and its integration into trading and FX platforms - gives banks a way to protect against currency volatility and offer clients better pricing. The next question is whether the model becomes an industry standard. Ants International has opened the model to developers via an API trial on GitHub, a move that could lower the barrier to high-end FX forecasting for smaller fintechs and e-commerce platforms.
Why this matters for finance professionals
As AI adoption in finance shifts beyond generative chatbots to specialized models built for time-series data, treasury teams will gain more predictability in cross-border operations. Falcon TST's core capability - turning raw transaction history into a forecast actionable enough to steer FX hedging and liquidity decisions - shows how AI can directly improve capital efficiency. The model's benchmark trial in public GitHub also suggests a roadmap: the same tool that powers decisions at the largest banks may soon be accessible to smaller teams for their own forecasting, a shift in how hedging and cash management are done across the Finance industry.
Your membership also unlocks:
