Vince Sitzmann: interaction-learned small models may hit the Hz needed for dexterous robot control
vincesitzmann · x · 2026-09-08
Responding to latency concerns, Vince Sitzmann argues LLMs and video models will become much faster — the tech is barely a decade old. Large models may not reach the hundreds of Hz needed for dexterous dynamic control, but smaller models actively learned via real-world interaction just might. He also speculates OpenAI may have added lots of robot and computer data to training, so some behavior may not be emergent.
More from Models
- DeepSeek spotted gray-testing two V4-Flash-Vision models, one faster but weaker — teortaxesTex · 2026-09-08
- GPT-6 Astra demos roundup: livestreams, retro games, 3D scenes and full websites — aitrendz_xyz · 2026-09-08
- DeepSeek V4.1 Flash beta goes live with new architecture, native multimodality, up to 507 tokens/s — 智东西 · 2026-09-08
- Grok 4.6 ranked #2 on AutomationBench-AA, beating Claude, GPT-5.6 and Kimi — XFreeze · 2026-09-08
- OpenAI API users report day-long outages: searches hang 30 minutes then drop with no output — msch6873 · 2026-09-08
- DeepSeek V4.1 Flash Survey Asks If It Can Fully Replace Production V4 Pro — zephyr_z9 · 2026-09-08