Post-Cutoff.com
  1. Home
  2. Timeline
  3. 2026
  4. AgiBot releases GO-2 embodied foundation model with action…

AgiBot releases GO-2 embodied foundation model with action chain-of-thought

★★★roboticsAgiBotconfidence: high

Shanghai's AgiBot released Genie Operator-2 (GO-2) on 2026-04-09, a VLA that plans in action space (action chain-of-thought) with an asynchronous slow-planner/fast-executor design; it reports 98.5% on LIBERO and 82.9% real-world success from simulation-only training.

Key facts

What happened

AgiBot, one of China's largest humanoid makers, followed its open GO-1 (March 2025) with GO-2, which tackles the gap between a model's reasoning and its motor execution.

Why it matters

Chinese humanoid makers are building their own robot foundation models, not just hardware.

Changelog

  • 2026-09-29: created

Models

Videos (1)

AGIBOT Unveils Genie Operator-2 (GO-2): Next-Gen Embodied Foundation Model

AGIBOT · 2026-04-09 · official

Description by Gemini, which watched the video:

Summary
This official demonstration video from AgiBot showcases GO-2 (Genie Operator-2), a general embodied foundation model controlling an AgiBot dual-arm humanoid robot. Operating at autonomous 1x speed, the robot demonstrates reasoning-driven manipulation (Action Chain-of-Thought / ACoT), dynamic multi-task execution with verbal user interruptions, and dexterous tool use resilient to human disturbance.

What is shown

  • Title and framework: Intro title cards introduce "GO-2 (Genie Operator-2) AGIBOT General Embodied Foundation Model" and "The Unity of Reasoning and Action" [00:00–00:04].
  • Table cleanup & dynamic replanning under verbal interruption [00:05–01:37]:
    • Prompt: "Clean the table and sort items by category. Then place the upper-left cup into the bowl." [00:06].
    • Visualized Action Chain-of-Thought (ACoT) decomposes perception and action steps [00:07–00:11].
    • The robot sorts toiletries into a bowl, hands over objects between grippers, and sets an upright bottle [00:12–00:44].
    • A user introduces spoken interruptions mid-task: "Place headphones in leather box" [00:45] and "My phone is missing, help me find it" [00:58]. The robot dynamically updates task queues, lifts a notepad to reveal the hidden phone [01:03], packs the headphones into the pouch [01:13], and finishes by nesting the cup inside the bowl [01:25–01:36].
  • Phone charging with dynamic disturbance recovery [01:38–02:50]:
    • Prompt: "Charge the phone. Plug the charger into the power outlet and connect the cable to the phone." [01:39].
    • A human moves the power block while the robot reaches for it; the robot relocalizes after disturbance [01:42–01:48].
    • Plugs the power adapter into an outlet strip requiring millimeter-level precision [01:52–02:01].
    • Picks up the smartphone with one hand while a human pulls the charging cable away; the robot re-tracks and grasps the connector [02:08–02:26].
    • Inserts the charging cable directly into the phone's port with dual-arm coordination, activating the charging screen [02:30–02:45].

Claims & numbers

  • Video specifies playback speed as autonomous real-time ("1x autonomous") [00:06, 01:39].
  • Onscreen caption claims "Millimeter-level precision manipulation" during plug and connector insertion [02:00, 02:32].

Notable quotes

  • [00:45] "Place headphones in leather box."
  • [00:58] "My phone is missing, help me find it."

Assessment
This is an official demonstration video presenting real-world autonomous robotic manipulation running at 1x speed. The demos cleanly illustrate dynamic task switching, visual-tactile relocalization after physical human interference, and fine bimanual insertion skills without cuts within the execution phases.

Described by gemini-3.8-flash on 2026-09-29 from the video's audio and frames.

Related events

  1. Unitree open-sources UnifoLM-WLA-1.0 humanoid foundation model (Apache-2.0) ★★★

Sources (3)

id: 2026-04-09-agibot-go-2 · updated 2026-09-29 · open in the interactive timeline