AW-0 runs at 230ms on a single RTX 5090, 2.4x faster than autoregressive code planning
trevordarrell · x · 2026-10-27
Part 3/8 of the AW-0 thread: planning without a separate autoregressive stage. AW-0 jointly denoises discrete code tokens, continuous video latents, and robot actions. On a single RTX 5090, AW-0 takes 230 ms vs 190 ms for a code-less WAM and 550 ms for autoregressive code—adding planning costs just 40 ms. Joint denoising is 2.4x faster than autoregressive code generation while winning 7 of 8 real-world tasks.
More from Embodied
- DONIC hits 86.79% on bimanual dexterous tracking, crushing prior SOTA's 9.43% — liuziwei7 · 2026-10-27
- Bedrock Robotics raises $600M as its fully autonomous excavators do paid work on live job sites — Sethwinterroth · 2026-10-27
- AW-0 hits 80.8% on RoboTwin tasks, beating strongest baseline by 16 points — trevordarrell · 2026-10-27
- AW-0's code stream fixes world model prediction: picks mango where baseline grabs apple — trevordarrell · 2026-10-27
- AW-0 debuts: one generative world action model plans in code, predicts, and controls robots — trevordarrell · 2026-10-27
- Skyfall AI's SKY-001 world model turns sketches into CAD builds in 3.9s, vs 199s for GPT-6 Astra — rohanpaul_ai · 2026-10-27