The Physical AI Marketplace
Apps for Reachy Mini, datasets and policies for LeRobot, and the tools an imitation-learning bench runs on — curated from the open ecosystem by the Hyperion lab.
Curated open-source from the Reachy Mini and LeRobot ecosystem, plus Hyperion's own artefacts where marked. Third-party items are not Hyperion products; each card links to its source and licence.
Apps for Reachy Mini
Official and community apps for the Reachy Mini desktop robot, plus its published move libraries — installed from Hugging Face Spaces.
reachy_mini_conversation_app
by pollen-roboticsThe official voice-conversation app for Reachy Mini, from its maker.
Reachy Minibaby-reachy-mini-companion
by ravediamondCommunity companion app that gives Reachy Mini a baby-like interactive personality.
Reachy Minireachy_mini_3d_web_viz
by 8bitkickCommunity 3D visualisation of Reachy Mini in the browser.
Reachy Minireachy_mini_language_tutor
by mattdotvaughnCommunity app that turns Reachy Mini into a language-practice tutor.
Reachy Minireachy_mini_home_assistant
by djhui5710Community bridge app connecting Reachy Mini to a Home Assistant installation.
Reachy Minireachy_mini_testbench
by pollen-roboticsThe official debug and test-bench app for Reachy Mini development.
Reachy Minireachy_mini_reactions
by 8bitkickCommunity app for on-device, voice-triggered expressive reactions.
Reachy Minireachy_mini_radio
by pollen-roboticsThe official radio-player app for Reachy Mini.
Reachy Minireachy_mini_remote_control_app
by pollen-roboticsThe official remote-control app for driving Reachy Mini by hand.
Reachy MiniReachy_Mini_Assembly_Guide
by pollen-roboticsThe official step-by-step assembly guide for the Reachy Mini kit.
Reachy Minireachy-mini-emotions-library
by pollen-roboticsCurated emotion trajectories for Reachy Mini — JSON moves (head, antennas, body yaw) with audio.
apache-2.0Reachy Minireachy-mini-dances-library
by pollen-roboticsCurated dance trajectories for Reachy Mini, maintained by Pollen Robotics.
apache-2.0Reachy Mini
LeRobot datasets & policies
Imitation-learning datasets and trained policies in LeRobot format — from benchmark sims to cross-embodiment sets covering SO-100/SO-101 arms.
pusht
by lerobotThe PushT manipulation benchmark in LeRobot format — 206 teleop episodes.
mitsim-onlydroid_1.0.1
by lerobotThe large-scale DROID Franka manipulation dataset in LeRobot v3 format — 95,658 episodes.
apache-2.0Frankaaloha_sim_transfer_cube_human
by lerobotALOHA bimanual simulation dataset — human-teleop cube transfer, 50 episodes.
mitsim-onlyaloha_sim_insertion_human
by lerobotALOHA bimanual simulation dataset — human-teleop peg insertion, 50 episodes.
mitsim-onlycommunity_dataset_v3
by lerobotCross-embodiment community pretraining dataset for VLA models — includes SO-100/SO-101 arms.
apache-2.0SO-100SO-101libero
by lerobotThe LIBERO benchmark suites in LeRobot format — Panda simulation, 1,693 episodes.
apache-2.0sim-onlysmolvla_base
by lerobotSmolVLA base — the compact open vision-language-action policy for LeRobot.
apache-2.0diffusion_pusht
by lerobotDiffusion Policy trained on PushT — the classic LeRobot dataset/policy baseline pair.
apache-2.0sim-onlyact_aloha_sim_transfer_cube_human
by lerobotACT policy trained on the ALOHA cube-transfer dataset.
apache-2.0sim-onlyxvla-base
by lerobotXVLA base vision-language-action policy in LeRobot format.
apache-2.0pi05_base
by lerobotπ0.5 base vision-language-action weights in LeRobot format (Gemma licence terms).
gemma
Tools & skills
The working stack: the LeRobot library, LeLab's browser workflows, and the canonical Reachy Mini app index.
LeRobot
by lerobotThe open robotics library behind SO-101-class arms — datasets, policies, training and evaluation.
apache-2.0SO-100SO-101Reachy MiniLeLab
by lerobotBrowser UI for LeRobot workflows — teleoperate, record and replay episodes from a local machine.
SO-100SO-101reachy-mini-official-app-store
by pollen-roboticsThe canonical index of Reachy Mini apps, maintained by Pollen Robotics.
Reachy Mini
Edge hardware: Strix Halo
The 128 GB unified-memory APU class the Hyperion bench runs on — AMD Ryzen AI Max+ 395 tooling for local inference: llama.cpp and vLLM toolboxes, hybrid NPU serving, benchmarks and setup guides.
AMD Ryzen AI Max+ 395
by AMDThe official spec sheet for the Strix Halo APU — 16 Zen 5 cores, Radeon 8060S (40 CU, RDNA 3.5), XDNA 2 NPU (50+ TOPS), up to 128 GB LPDDR5X-8000 unified memory.
Strix HaloStrix Halo AI Toolboxes
by kyuz0The community hub for running local LLMs on Strix Halo — containerised llama.cpp environments per backend, kept current with upstream.
Strix Haloamd-strix-halo-toolboxes
by kyuz0Pre-built llama.cpp toolbox containers for gfx1151 — ROCm and Vulkan backends, rebuilt as llama.cpp master moves.
Strix Haloamd-strix-halo-vllm-toolboxes
by kyuz0Toolbx-compatible containers for serving LLMs with vLLM on Strix Halo (gfx1151).
Strix HaloLemonade
by lemonade-sdkOpen-source local LLM server with AMD optimisations — hybrid execution across the XDNA 2 NPU and Radeon GPU on Ryzen AI machines, OpenAI-compatible API.
Strix HaloRyzen AI 300Strix Halo Wiki
by communityCommunity wiki for local AI on Strix Halo — setup guides including llama.cpp with ROCm.
Strix Halo
AI agents
The software-agent side of the marketplace — specialised chat agents, catalogue and team workflows.