LitXBench (Coding Agents) - Measurements: leaderboard

LitXBench (2026) tests extracting complete experiments from scientific papers: on LitXAlloy, 19 alloy papers with 1,426 annotated measurements, the model writes each material with its processing chain, microstructure and measurements as Python objects from the OCR text of the paper. F1 from 0 to 1 on the extracted measurement values, for coding agents (Claude Code, Codex, Gemini CLI); higher is better.

Metric: Measurement F1 (0-1), extracted measurement values against the annotated ones (kind, value, unit, uncertainty, temperature and pressure) for extracting every experiment (materials, process chains, microstructure configurations and measurements) as Python objects from the OCR text of LitXAlloy's 19 alloy papers (1,426 measurements; figures excluded), extracted materials matched to the annotated ones by the Hungarian algorithm, mean of three runs; coding agents (Claude Code, Codex, Gemini CLI); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 3 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3.1 Pro (Preview)0.74#99
2Claude Opus 4.60.71#113
3GPT-5.2 CodexHigh0.67#182

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=litxbench-coding-agents-measurements · How It Works · Data refreshed daily, snapshot 2026-11-21.