🍳 RoboCasa Kitchen Leaderboard
A neutral, community-maintained leaderboard for Vision-Language-Action (VLA) policies — by GINIGEN-AI
📖 What is this?
RoboCasa Kitchen is a robot-learning benchmark where a single-arm robot (Franka Panda) performs 24 atomic kitchen manipulation tasks — picking objects, opening drawers/doors, turning faucets, pressing buttons, and more — inside a photorealistic simulator.
This board tracks the average success rate of Vision-Language-Action (VLA) policies: models that take camera images + a natural-language instruction as input and output low-level robot actions. It is the bridge between large multimodal models and real embodied control.
RoboCasa has no official leaderboard, and published numbers are scattered across papers under different protocols. GINIGEN-AI aggregates them neutrally, compares within a single matched protocol (the RLDX-1 Technical Report — the most recent head-to-head comparison), and cites every source.
🏆 Kitchen 24-task (matched protocol)
| # | Model | SR (%) | Org | Backbone/Arch | Paper | Date | Protocol |
|---|---|---|---|---|---|---|---|
| 1 | RLDX-1 | 70.6 | RLWRLD / KAIST (Jinwoo Shin) | MSAT | arXiv:2605.03269 | 2026-05 | matched (RLDX-1 report) |
| 2 | GR00T N1.6 | 66.2 | NVIDIA | GR00T | arXiv:2503.14734 | 2026 | reproduced in RLDX-1 |
| 3 | GR00T N1.5 | 65.7 | NVIDIA | GR00T | arXiv:2503.14734 | 2025 | reproduced in RLDX-1 |
| 4 | π0-FAST | 63.6 | Physical Intelligence | π0 (FAST tok.) | — | 2025 | reproduced in RLDX-1 |
| 5 | π0 | 62.5 | Physical Intelligence | π0 (flow) | arXiv:2410.24164 | 2024 | reproduced in RLDX-1 |
| 6 | π0.5 | 62.1 | Physical Intelligence | π0.5 | arXiv:2504.16054 | 2025 | reproduced in RLDX-1 |
➕ Other protocols (not directly comparable)
| Model | Reported SR (%) | Org | Backbone | Source | Why separate |
|---|---|---|---|---|---|
| Cosmos-Policy (Predict2-2B) | 67.1 | NVIDIA | 2B (video-FM finetune) | model card | 50 demos, NVIDIA-own protocol |
🤖 GR1-Tabletop variant (separate suite)
| Model | GR1-Tabletop 24-task SR (%) | Org | Source |
|---|---|---|---|
| GR00T (h=16) | 59.7 | NVIDIA | arXiv:2604.05014 (StarVLA) |
| TwinBrainVLA | 54.6 | — | arXiv:2601.14133 |
| VP-VLA | 53.8 | — | arXiv:2603.22003 |
| StarVLA-OFT | 48.8 | — | arXiv:2604.05014 |
| GR00T-N1.6 | 24.2 | NVIDIA | RoboCasa-GR1 (no-collision 7.0%) |
📝 Submit your result
Submissions are stored privately and reviewed before appearing on the board.
Maintained by GINIGEN-AI. Not affiliated with the RoboCasa authors. Corrections welcome via the Community tab.