[KongchangAI]
BenchmarkTerminal Bench Science 0.1

Terminal Bench Science 0.1

A terminal-operation benchmark aimed at scientific research scenarios, requiring models to complete multi-step tasks related to scientific computing in a simulated Linux terminal environment, assessing command-line proficiency, long-horizon task planning and error-recovery ability.

Timeline (last 90 days)

Nov 4

Judging from the 12 benchmark scores compiled by a Bilibili uploader, Hy4 is basically ahead of or on par with DeepSeek and Kimi, falling significantly behind K3.1 Pro only in Terminal Bench Science 0.1.

Expired50%
Nov 4

Terminal Bench Science 0.1 is a terminal-operation benchmark aimed at scientific research scenarios, requiring models to complete multi-step tasks related to scientific computing in a simulated Linux terminal environment.

Unverified50%
Sep 4

Fable 5.1 scored 52.6% on the Terminal Bench Science 0.1 benchmark, more than doubling the score of its previous generation.

Unverified60%

All Facts (2)

Source Articles