---
title: VLABench
description: "Fudan OpenMOSS language-conditioned manipulation bench: 100 categories, 2,000+ objects in MuJoCo. WorldFoundry imports official rollouts; it does not run the sim."
---

VLABench (OpenMOSS, Fudan University) benchmarks **language-conditioned manipulation that needs world knowledge and multi-step reasoning**, not just instruction following. Official size: **100 task categories** and **2,000+ objects** in MuJoCo, probing mesh / texture understanding, spatial and semantic reasoning, physical-law understanding, and long-horizon planning for both VLA policies and VLM planners.

Paper: [VLABench](https://arxiv.org/abs/2502.09858). Code: [OpenMOSS/VLABench](https://github.com/OpenMOSS/VLABench). License: upstream terms.

WorldFoundry is an **official-results importer**. Simulator execution, policy loading, and rollout collection stay upstream. Scorecard primary: `success_rate`.

<BenchmarkPageSections id="vlabench" locale="en" />

## How to run

```bash
worldfoundry-eval zoo benchmark-run \
  --benchmark-id vlabench \
  --mode official-validation \
  --official-results-path <official_results.json> \
  --output-dir tmp/vlabench/official-validation \
  --json
```

Default import: `WORLDFOUNDRY_VLABENCH_RESULTS_PATH`. Optional root: `WORLDFOUNDRY_VLABENCH_ROOT`.

In-tree pointer: `worldfoundry/data/benchmarks/tasks/external/vlabench.yaml`.

Outputs include `scorecard.json` (primary `success_rate`), `raw_results.jsonl`, and `evaluation/summary.json`.

[← Benchmark Hub](/docs/evaluation/benchmark-hub)