UT Austin

RIPT-VLA↗

RIPT-VLA is a VLA from UT Austin, released May 22, 2025. It has no overall rank on the Robot Intelligence Index, which needs both a simulation and a real-robot result. Its best result is #3 of 11 on LIBERO-Plus. Interactive RL post-training (RIPT) applied to VLAs such as QueST and OpenVLA-OFT using sparse success rewards.

Release date: May 22, 2025

Developer
UT Austin
Type
VLA
Model weights
Open
License
MIT
Parameters
—
Pricing
Self-hosted
Benchmarks
3
Overall rank
Not ranked
Needs a real-robot result to rank overall.
Real robots rank
Not ranked
No real-robot results yet.
Simulation rank
#26of 49
Rating 1365 · range 19–32
Rating
1419± 22.9
10601989
Cost / task · est.
$0.0099
$0.0087$17.74
Time / task · est.
11s
9.5 s1 h 22 min
Rank on each benchmark
BenchmarkScoreRank
LIBERO-PlusOverall68.43 / 11Paper-reportedLIBERO avg93.6463 / 823LIBERO-Pro2Unperturbed57.314 / 20

Per task attempt; the tiles show the median across benchmarks. Self-hosted compute is priced at one H100 at $3.29 per hour (Lambda on-demand, Oct 2026).

Inference

RL post-training method; inference cost equals the base policy (QueST or OpenVLA-OFT).