← Back to leaderboard

llama3.2:3b

3.1 GB measured VRAM · sub-4GB tier · reasoning-gauntlet-v1
Exact problems, puzzles, and expected answers for this category are held back from publication to avoid training-data contamination — see About for why. This page shows per-tier pass/fail counts and timing only.
Tier 1 — floor 5/5
115 tokens · 7.215s (representative seed)
Tier 2 — multi-step 5/5
152 tokens · 0.716s (representative seed)
Tier 3 — logical deduction 5/5
238 tokens · 1.074s (representative seed)
Tier 4 — adversarial (CRT trap) 5/5
139 tokens · 0.719s (representative seed)
Tier 5 — combinatorics 5/5
270 tokens · 1.223s (representative seed)
Tier 6 — distractor filtering 5/5
152 tokens · 0.778s (representative seed)