AI Paper Atlas

Research map / Large language models / Agents, code and math

MolLangBench: A Comprehensive Benchmark for Language-Prompted Molecular Structure Recognition, Editing, and Generation

Feiyang Cai, Jiahui Bai, Tao Tang, Guijuan He, Joshua Luo, Tianyu Zhu, Srikanth Pilla, Gang Li, Ling Liu, Feng Luo

ICLR 2026 · Large language models › Agents, code and math · approach: LLMs for scientific discovery · 11 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use GPT-5 instead of o3LLM · MolLangBench structure recognitionaverage recognition accuracy: 0.877 -> 0.923Improved
use image input instead of SMILES texto4-mini · MolLangBench structure recognitionrecognition accuracy: 0.772 vs o3-mini-level language resultsHurt
require atom localization in addition to recognitionLLMs · MolLangBench structure recognitionaccuracy: GPT-5 0.923 -> 0.862Hurt
use GPT-5 on bond stereo (E/Z)LLMs · MolLangBench bond stereorecognition accuracy: GPT-5 0.655; o3 0.480; GPT-4o 0.390Mixed
use SELFIES instead of SMILES (ablation)LLMs · MolLangBenchaccuracyHurt
describe molecular structure before predicting property (ablation)GPT-4o, GPT-4.1 · BBBP, BACEaccuracy: improvement exceeds 5%Improved
vary chemical terminology or instruction phrasing (ablation)o3 · MolLangBench editing/generationaccuracyNo clear effect
use image generation model instead of text LLM for editingGPT Image 1 · MolLangBench molecule editingediting accuracy: 0.855 (GPT-5) vs 0.080Hurt
generate de novo instead of editingGPT-5 · MolLangBenchSMILES validity / accuracy: editing 0.945/0.855 -> generation 0.690/0.430Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy