pandaily

Robotera's VPP2 World Action Model Tops RoboDojo, Scores 58.5% Zero-Shot on Real ALOHA Arms

Robotera's open-source VPP2 tops RoboDojo's 42-task dual-arm benchmark at a 32.26% success rate by training video prediction first and robot actions second. Robotera, the Tsinghua University-backed ...
Pandaily

ByteDance's TRAE Merges TraeWork and TraeCode Into One Platform; TraeWork Shuts Down October 23

ByteDance's TRAE folds TraeWork and TraeCode into one app with Agent and IDE modes across desktop, web and mobile. TraeWork ...
pandaily

Galbot and Tsinghua's LATENT Wins an IROS Award for a Humanoid Tennis Forehand

Want to read in a language you're more familiar with? LATENT, from Galbot, Tsinghua and collaborators, won an IROS 2026 entertainment paper award. In the paper, a Unitree G1 returned forehands on ...
Pandaily

ByteDance's Doubao App Lands on Huawei HarmonyOS PCs With Doubao-Seed-2.1 Pro and Turbo

Doubao's native HarmonyOS PC app ships with Doubao-Seed-2.1 Pro and Turbo, chat and work modes and agent-style tools; Huawei ...
Pandaily

Chinese Team Maps Rice's Full Life Cycle in 3D Cell Atlas Published in Cell

A Chinese-led team has published in Cell what it calls the first three-dimensional spatiotemporal cell atlas spanning rice's ...
Pandaily

DeepSeek Harness v0.2.1-alpha.1 Adds Experimental Claude Code Mods Compatibility Layer

DeepSeek Harness v0.2.1-alpha.1 adds an experimental bridge that runs Claude Code Mods as Harness plugins, plus a new entry ...
Pandaily

China's BEST Fusion Project Campus Delivered Early as Main Machine Enters Four-Ring Assembly

The campus of BEST, China's compact fusion experiment in Hefei, was delivered early. The main machine has begun four-ring ...
pandaily

Horizon Robotics Releases HSD V2.1 With End-to-End Reversing, iCAR V27 First in Line for OTA

Horizon Robotics' HSD V2.1 adds end-to-end reversing in all scenarios, with 43% more miles per intervention and 255% better interactive lane changes. iCAR V27 OTA starts October 8. Horizon Robotics on ...
pandaily

StepFun Step 5 Preview: ~600B MoE Long-Horizon Agent; Open Weights Due Oct 15

StepFun launched Step 5 Preview on September 20, 2026: a ~600B sparse MoE (~27B active) with 1M context for long-horizon agents, API live now and BF16 open weights planned for October 15. StepFun ...
pandaily

StepFun Launches Step 5 Preview: 600B Sparse MoE, 1M Context, Weights Open Oct 15

StepFun’s Step 5 Preview pairs a 600B/27B sparse MoE with a 92-layer narrow-deep stack and 1M context for long-horizon agents, with weights planned for October 15. StepFun has launched Step 5 Preview, ...
pandaily

Tencent Open-Sources WeKnora With Official DeepSeek Harness Plugin

Want to read in a language you're more familiar with? Tencent WeChat open-sources WeKnora, a RAG-to-agent knowledge platform with 25k+ GitHub stars, plus an official DeepSeek Harness plugin that adds ...
pandaily

Shanghai AI Lab Releases Atria Dawn Preview, a 744B MoE Agentic Model

Want to read in a language you're more familiar with? Shanghai Artificial Intelligence Laboratory open-sourced Atria Dawn Preview under MIT, a 744B MoE agentic model built on GLM-5.2 with 256K context ...