PatronusAI(@PatronusAI):Today, we’re releasing SpeedrunBench: the first benchmark that measures how fast agents can beat video games. We asked frontier models not to simply complete video games, but to speedrun them - across 10 titles like Mario Kart and Pokemon. On simpler games, agents get close to the human world record. However, on more complicated games like Pokemon Blue, the best models Kimi-K3 and Opus 5 are ~4x off the world record. Benchmark, paper, and demo below.

PatronusAI

@PatronusAI

Simulation research and infrastructure for human-aligned AGI

加入 July 2023

220 正在关注 2.6K 粉丝

PatronusAI@PatronusAI

2026.09.03 19:20

Today, we’re releasing SpeedrunBench: the first benchmark that measures how fast agents can beat video games. We asked frontier models not to simply complete video games, but to speedrun them - across 10 titles like Mario Kart and Pokemon. On simpler games, agents get close to the human world record. However, on more complicated games like Pokemon Blue, the best models Kimi-K3 and Opus 5 are ~4x off the world record. Benchmark, paper, and demo below.

显示更多