Gemini 3.7 Flash scores 92% on robot tool benchmark, marking a capability leap
minsuk_chang · x · 2026-08-18
Gemini 3.7 Flash achieved a 92% score on a physical tool-use benchmark, effectively saturating it. In contrast, Gemini 3.6 Flash, released just three weeks prior, scored only 32%. This dramatic improvement marks a step change in the robotics capabilities of LLMs.
More from Embodied
- Noble Machines named to BI's 2026 list of top robotics startups — ditzikow · 2026-08-18
- Dyna Robotics: training on 1M hours of egocentric video broke everything that worked at 10K — ZeYanjie · 2026-08-18
- Unitree previews Superman humanoid that outjumps humans and outsprints Usain Bolt — GraceToSentience · 2026-08-18
- Uber invests in Zipline, scaling autonomous deliveries to over 1M daily — AnandSwa · 2026-08-18
- Lamp hardware features local agent runtime for autonomous thinking — dee_hw · 2026-08-18
- Aheadform unveils bionic humanoid Elf Xuan 2.0 with metal body at WRC — CyberRobooo · 2026-08-18