Skip to content

VLA-JEPA on the Orin for $9.7, and why best-of-N reranking is a no-go

We wanted to know if picking the best of several planned motions could improve a robot policy. We sampled 8 motion chunks per step and measured how much they differed and what they cost. They barely differed, took 1.84× the time, and their plain average captured 45 % of the possible gain.

Time per chunk, relative to 1 sample (lower is better): 1 sample 1.00×; 8 samples 1.84×; bar 1.5×1 sample1.00×8 samples1.84×bar 1.5×

Evaluation and judges · 11 of 14

Type: engineering · Result: VLA-JEPA fine-tuned for ~$9.7 scores 9/10 on the arm; best-of-8 reranking is a no-go (0.76, 1.84×). · n: 10 trials; H0 on the teleop val split

Not proven: The paper's robustness claim is not tested.

Compute: External GPU: yes (vast.ai): ~$9.7 for the fine-tune (retries not itemised). H0 and the arm gate on the Orin NX: $0. · Reproduce: 2026-10-23-paper-vs-hardware-vla-jepa · First published: 2026-10-08 · Updated: 2026-10-08

The first world-model idea was the obvious one: let the policy propose several action chunks, let a latent world model imagine each one, keep the best. Before building any of it, I checked whether the premise held. It did not.

VLA-JEPA on the arm

Fine-tuned on vast.ai for about $9.7 (over the $3.5–4.7 estimate: interruptible bids cost more than they save on long runs). On the Orin NX, world model off at inference: 9/10 on bowls, 355 ms per chunk. Same score as SmolVLA at three times the training cost. The task is at its ceiling and cannot tell them apart; that comparison waits for a harder primitive.

H0: do the samples vary enough to pick from?

Pre-registered bars: oracle-to-mean error ratio ≤ 0.7, latency for 8 samples ≤ 1.5× one sample, baseline success below 85 %.

  • The 8 samples are bundled: 1.08° spread, oracle/mean ratio 0.76 (null 0.92).
  • Latency 1.84× (643 vs 349 ms).
  • Baseline success 90 %, already above the bar where reranking could pay.
  • Averaging the 8 chunks, with no model at all, captures 45 % of the oracle gain.

Verdict: no-go. A reranker cannot fix a bias the samples share, and the cheapest competitor (the mean) takes half the available gain for free.

What changed

Reranking left the architecture the same day. The question became whether a world model can see a failure, not whether it can choose an action. That turned out to be a much better question, with a much longer answer.

Compute. External GPU: yes (vast.ai): ~$9.7 for the fine-tune (retries not itemised). H0 and the arm gate on the Orin NX: $0.
Reproduce this: 2026-10-23-paper-vs-hardware-vla-jepa

Evaluation and judges, 11 of 14. ← Evaluation and judges 10: TurboVLA: a policy that ignored its instruction · Evaluation and judges 12: Audit your labels where they matter: a VLM labeller that is right on average →