SYLIN

Terminal-Bench

Benchmark · Text & code

Commenter says Terminal-Bench 2.1 is saturated and wants AA to move to 4.0, which differentiates models better

Versions

VersionReleasedRank
Terminal-Bench 4LATEST
Terminal-Bench 2.1

Evidence

  1. Sep 3
    Terminal-Bench Science: Fable 5.1 scores 52.6%, over double most other models
  2. Jul 22
    Kimi K3 launch table shows K3 ahead of Fable 5 on Terminal-Bench (self-reported)

Alternatives

GLMOn agentic coding benchmarks incl. Terminal-Bench, Agents' Last Exam, AutomationBench, GDPval it sits at or near the top of open models, near closed frontier
Made for Terminal-Bench