Search results

Found 2 results for "ibm research"

ScarfBench: New Benchmark Tests AI Agents on Enterprise Java Migration

IBM Research launches ScarfBench, an open benchmark evaluating AI agents on cross-framework Java migration, revealing configuration and runtime dependencies as top hurdles.

Maya Bennett·Jul 29, 2026

AI & Software

Why Model Routing Is Harder Than It Looks: Lessons from IBM Research

IBM Research found that routing models in production is a systems optimization problem involving cost, latency, and complexity beyond simple task difficulty.

Maya Bennett·Jul 18, 2026