0

Physgym Arena Drhard Public

PhysGym Arena DR-hard benchmark for domain-randomized Gym simulator repair

Domain
rl-env
License
unknown
Published
May 2026

Cite

Notes

Only stored in your browser.

Attribution

Leaderboard scores
prime-hub
Attribution policy →

Top score 45.3% by Claude Opus 4.7 - 5 models reporting (2 frontier)

Score history

5
40%55%70%85%100%Sep 25Nov 25Jan 26Mar 26Qwen3 MaxClaude Opus 4.7

Top models

5
Physgym Arena Drhard PublicBar chart with 5 bars. Highest value: Claude Opus 4.7 at 45.3.
5 models

Related tools

1
View all

Implementations, trainers, datasets and scaffolds linked to this eval.

FAQ

What is Physgym Arena Drhard Public?
PhysGym Arena DR-hard benchmark for domain-randomized Gym simulator repair
What is the current top score on Physgym Arena Drhard Public?
The top reported score is 45.3% by Claude Opus 4.7, across 5 models reporting (2 from frontier labs).
How can a model improve its Physgym Arena Drhard Public score?
Tools linked to Physgym Arena Drhard Public on Sophon include Drhard Public RL Env (Community) - RL environments, datasets, and scaffolds that target this eval.
What license is Physgym Arena Drhard Public under?
Physgym Arena Drhard Public is available under unknown.