# OpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory > Source: > Published: 2026-10-20 04:30:01+00:00 Multimodal large language models (MLLMs) have made rapid progress on visual reasoning, yet many tasks require evidence that is difficult to obtain through internal reasoning alone: our controlled study on 22 benchmarks shows that simply increasing the inference budget of the base model yields only m