MAVEN: Improving Generalization in Agentic Tool Calling

Omkar Ghugarkar, Vishvesh Bhat, Muhammad Ahmed Mohsin, Asad Aali · 2026

Read MAVEN: Improving Generalization in Agentic Tool Calling by Omkar Ghugarkar, Vishvesh Bhat, Muhammad Ahmed Mohsin, Asad Aali (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-30738.txt.

Description

Ghugarkar and colleagues present MAVEN, a scaffold for decomposition, orchestration, and verification in tool calling. Large language models score strongly on benchmarks but struggle to compose strategies, preserve states, and coordinate tools across domains. MAVEN is evaluated across tool-calling benchmarks, plus MAVEN-Bench, a test for multistep mathematical and physical reasoning. MAVEN-Bench exposes a gap between quality and end-to-end success. In runs, MAVEN lifts its base model from 48 percent to 71 percent accuracy without training, staying competitive at a tenth of the cost.

Catalog record

CREATOROmkar Ghugarkar, Vishvesh Bhat, Muhammad Ahmed Mohsin, Asad Aali
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30270
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.30738

Browse the full library · About the Tower of Babel