yuu-sim-1.1 × coding-depth@1.0.0

completed

Score

55.6%

95% CI 22%–89%

Pass rate

56%

3 tasks · 9 results

Cost

$0.00034

660 tok

p50 / p95

1.9s

p95 2.7s

Errors

unmeasurable / provider

Seed

mock

Provenance

benchmark hash: 3766b7e00cec3321

arena version: 1.0.0

node: v24.19.0

sandbox: vm-worker

live providers: no (mock)

repeats: 3

temperature: 0.2

ablations: none

By judge type

Deterministic and model-graded scores are never blended.

deterministic

56%

By capability

Coding56%

Failure taxonomy

Automatic classification; a human can override any label.

Refusal · 2Wrong answer · 1Coding runtime error · 1

Failure clusters

Slices that fail far more than the rest of the run.

tagged “cache”+83 pts

100% failure on tagged “cache” (6/6), 83 points worse than elsewhere; mostly wrong answer.

hard difficulty+83 pts

100% failure on hard difficulty (3/3), 83 points worse than elsewhere; mostly wrong answer.

tagged “data-structure”+83 pts

100% failure on tagged “data-structure” (3/3), 83 points worse than elsewhere; mostly wrong answer.

Trace viewer · 9 results

Every task attempt, with the exact input, raw output, tool calls, and score. Click to expand.

Runs/