Codesota · Benchmark · RLBenchHome/Leaderboards/RLBench
Imperial College London

RLBench.

Large-scale robot learning benchmark with 100 diverse manipulation tasks in simulation. Standard multi-task benchmark for language-conditioned robotic manipulation. Evaluated on 18 tasks with 100 demonstrations.

Paper ↗Leaderboard ↓
§ 01 · Leaderboard

Results by metric.

Only 3 models on this benchmark
Help build the community leaderboard — submit your model results.
Found a wrong score or missing run?
Use row edits to send a sourced correction into moderation.
Add / edit result ↗Report issue ↗

Success Rate (%)

Average task success rate across 18 RLBench manipulation tasks with 100 demonstrations each.

Higher is better

Trust tiers for Success Rate (%)verifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

RankModelTrustScoreYearLinksFix
01RVT-2
Average success rate across 18 RLBench tasks (100 demos). Table I in paper. NVIDIA, June 2024.
verified81.42026Source ↗Looks wrong?
02RVT
Average success rate across 18 RLBench tasks (100 demos). Reported as baseline in RVT-2 paper, Table I.
verified62.92026Source ↗Looks wrong?
03PerAct
Average success rate across 18 RLBench tasks (100 demos). Table 1 in paper. CoRL 2022.
verified43.42026Source ↗Looks wrong?
§ 04 · Submit a result

Add to the leaderboard.

← Back to Leaderboards