The bank's AI foundation moved from proofs of concept to production in code development, research automation, and AML compliance, with more than 15,000 developers using AI coding assistants.
The 8.79B model is designed to beat larger systems on math and code while using shorter reasoning traces, but its non-commercial license and reliance on IBM Granite shape its deployment path.
The Oct. 17 build of DeepSeek-V4.1-Flash improved 11 benchmarks and shortened outputs while `deepseek-flash` kept its model ID and pricing, resetting expectations for how frontier-model updates reach developers.
CEO Brian Moynihan put a 2-to-1 return on the record at a Barclays conference and said the bank's 2027 AI budget will double.
Recommended reading
The field-tested playbook for leading AI transformation — from literacy and governance to ROI, security, and scale.
Get the book on Amazon (opens in a new tab)Paid link. As an Amazon Associate, we earn from qualifying purchases.