The authors propose an enhanced version of the van der Waals density functional (vdW-DF2), which employs a more accurate semilocal exchange functional (PW86) and a large-$N$ asymptote gradient correction to improve the prediction of binding energy, equilibrium separation, and potential-energy curve shape. This new functional is compared with accurate quantum chemical calculations on 22 duplexes, showing significant improvements in equilibrium separations, hydrogen bond strengths, and vdW attractions at intermediate separations. The vdW-DF2 functional is expected to enable chemically accurate calculations in sparse materials, which are crucial for understanding condensed matter, surface phenomena, and biological systems. The method is validated through comparisons with reference datasets and extended solid systems, demonstrating its potential for applications in various fields.The authors propose an enhanced version of the van der Waals density functional (vdW-DF2), which employs a more accurate semilocal exchange functional (PW86) and a large-$N$ asymptote gradient correction to improve the prediction of binding energy, equilibrium separation, and potential-energy curve shape. This new functional is compared with accurate quantum chemical calculations on 22 duplexes, showing significant improvements in equilibrium separations, hydrogen bond strengths, and vdW attractions at intermediate separations. The vdW-DF2 functional is expected to enable chemically accurate calculations in sparse materials, which are crucial for understanding condensed matter, surface phenomena, and biological systems. The method is validated through comparisons with reference datasets and extended solid systems, demonstrating its potential for applications in various fields.