KYBER 0.1.0 · gate GREEN · fp 1adbc0d3 · 1,500 trials

The measured field · 01 / 25

The KYBER leaderboard.

Actual local-model strikes. One current, complete suite per model. Pass@1 decides the order; grade-weighted pass breaks ties. Every score links back to its run contract and trial evidence.
4 / 25published

Suite

1,500 gated trials

Ranking

pass@1 · full suite only

The podium

4 qualifying runs

02 · measured

34.3%

lfm2.5-2.6b-fable5-coding-agent-heretic

weighted 0.348 · 1500 trials

Read the report
03 · measured

27.8%

minicpm5-2b-claude-fable5-1-thinking-agentic

weighted 0.283 · 1500 trials

Read the report

Current · full suite

Top 25 results

4 real results qualify today. The board expands as new complete strikes are synced; no scores are simulated to fill the remaining positions.

Wilson uses trial rows; independent answer-shape clusters are shown on each model profile.

Inspect run provenance