TurboVLA: 0.2B Parameter Model Achieves 32Hz Robotic Inference on RTX 4090
_akhaliq · x · 2026-07-31
Current Vision-Language-Action (VLA) models typically adopt an LLM-centric V→L→A pathway, incurring massive computation and memory overhead at every inference step.
The paper introduces TurboVLA, which reformulates this pathway into a direct V+L→A mapping. Instead of using an LLM as the central interface between perception and action, TurboVLA encodes visual and linguistic instructions independently and predicts continuous action chunks via lightweight bidirectional interaction.
On the LIBERO benchmark, TurboVLA achieves a 97.7% average success rate with only 0.2B parameters. It runs on a consumer-grade RTX 4090 with a 31.2 ms inference latency and less than 1 GB VRAM, matching or outperforming significantly larger VLA policies.
Related event: TurboVLA Enables 32Hz Robot Control on Consumer GPUs(2 posts)→
More from Embodied
- List of 15 best robotics researchers on X, from Jim Fan to Chelsea Finn and Fei-Fei Li — FinanceYF5 · 2026-10-16
- Fake Apple listing for dual-screen 'iPhone Duo' jokes it's the best phone for watching your AI agents — AIFlow_ML · 2026-10-16
- Inside Auki's Hong Kong office: a fake grocery store powering robotics data collection — LoganGrasby · 2026-10-16
- Physicist calls compressed-air car a scam: one charge would take a week of compressor time — rickasaurus · 2026-10-16
- Robot gives thumbs-up from inside an upside-down car: 'peak tech moment' — NicoVerderosa · 2026-10-16
- NVIDIA launches RTX Spark for local AI on Windows PCs — ostrisai · 2026-10-16