0

Infraresolutionbench

Frontier

Prime verifiers environment for InfraResolutionBench

Domain
rl-env
License
unknown
Published
Apr 2026

Cite

Notes

Only stored in your browser.

Attribution

Leaderboard scores
prime-hub
Attribution policy →

Top score 92.2% by Claude Sonnet 4.6 - 35 models reporting (16 frontier)

Score history

32
55%66%78%89%100%Jul 25Sep 25Nov 25Jan 26Mar 26Grok 4GPT-5Claude Sonnet 4.5Claude Opus 4.6Claude Sonnet 4.6

Top models

35
InfraresolutionbenchBar chart with 21 bars. Highest value: Claude Sonnet 4.6 at 92.2.
21 models

Related tools

1
View all

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

FAQ

What is Infraresolutionbench?
Prime verifiers environment for InfraResolutionBench
What is the current top score on Infraresolutionbench?
The top reported score is 92.2% by Claude Sonnet 4.6, across 35 models reporting (16 from frontier labs).
How can a model improve its Infraresolutionbench score?
Tools linked to Infraresolutionbench on Sophon include Infraresolutionbench RL Env (Community) - RL environments, datasets, and scaffolds that target this eval.
What license is Infraresolutionbench under?
Infraresolutionbench is available under unknown.