News confirmed high confidence

Ant Group's Robbyant open-sources LingBot-VLA 2.0 (plus LingBot-World 2.0, LingBot-VA 2.0, LingBot-Vision)

Robbyant (Ant Group) upgraded and open-sourced LingBot-VLA 2.0, a 6B cross-embodiment VLA model pre-trained on 60,000 hours of real-world physical data (50k robot + 10k first-person human), spanning 20 morphologies from 17 manufacturers; ou

On 2026-07-07, the verified AI news record added a significant robotics, science & health development: Robbyant (Ant Group) upgraded and open-sourced LingBot-VLA 2.0, a 6B cross-embodiment VLA model pre-trained on 60,000 hours of real-world physical data (50k robot + 10k first-person human), spanning 20 morphologies from 17 manufacturers; outperformed π0.5 and GR00T N1.7 on the GM-100 dual-arm benchmark. Companion releases: LingBot-World 2.0 (hour-long 720p/60fps interactive world model, Jul 9), LingBot-VA 2.0 (93.6% RoboTwin-2.0 success; latency 927→142 ms, Jul 10), LingBot-Vision/Depth 2.0 (Jul 7).

Context

Robbyant (Ant Group) upgraded and open-sourced LingBot-VLA 2.0, a 6B cross-embodiment VLA model pre-trained on 60,000 hours of real-world physical data (50k robot + 10k first-person human), spanning 20 morphologies from 17 manufacturers; outperformed π0.5 and GR00T N1.7 on the GM-100 dual-arm benchmark. Companion releases: LingBot-World 2.0 (hour-long 720p/60fps interactive world model, Jul 9), LingBot-VA 2.0 (93.6% RoboTwin-2.0 success; latency 927→142 ms, Jul 10), LingBot-Vision/Depth 2.0 (Jul 7). Biggest open-weight embodied-model drop of the window; a six-part open toolkit positioning Ant as the "Android" layer of Chinese embodied AI.

What changed

Robbyant (Ant Group) upgraded and open-sourced LingBot-VLA 2.0, a 6B cross-embodiment VLA model pre-trained on 60,000 hours of real-world physical data (50k robot + 10k first-person human), spanning 20 morphologies from 17 manufacturers; outperformed π0.5 and GR00T N1.7 on the GM-100 dual-arm benchmark. Companion releases: LingBot-World 2.0 (hour-long 720p/60fps interactive world model, Jul 9), LingBot-VA 2.0 (93.6% RoboTwin-2.0 success; latency 927→142 ms, Jul 10), LingBot-Vision/Depth 2.0 (Jul 7). According to roboticstomorrow.com, the supporting record states: “LingBot-VLA 2.0 was pre-trained on 60,000 hours of high-quality, real-world physical data… Sourced from 20 distinct robot morphologies across 17 leading manufacturers—including Leju, AgiBot, Unitree, AgileX, Galaxea… Franka… Fourier… and Qinglong.”.

Why it matters

Biggest open-weight embodied-model drop of the window; a six-part open toolkit positioning Ant as the "Android" layer of Chinese embodied AI. The applied-AI angle matters because physical and scientific deployments test whether models can produce reliable outcomes outside chat and coding environments.

Details

The research file records the item under “Ant Group's Robbyant open-sources LingBot-VLA 2.0 (plus LingBot-World 2.0, LingBot-VA 2.0, LingBot-Vision)” with source timing of Jul 7–11, 2026. The captured research confidence note is: High. Additional captured source links are listed below so readers can inspect the evidence trail rather than rely on a single summary. Biggest open-weight embodied-model drop of the window; a six-part open toolkit positioning Ant as the "Android" layer of Chinese embodied AI.

Limitations and caveats

The research file did not identify a blocking caveat, but vendor-supplied claims should still be read as company statements unless independently confirmed.

Sources

Update note: Last reviewed 2026-07-22. Next checkpoint: monitor official channels and the linked source record.

Sources

Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.