Spaces:

---

bench-labs

/

BenchLabs-Leaderboard

like 9

Running

Adds cagliostro-v1 (157.4M) to the leaderboard.

Generated by script.py at sha256 aaec13fe93213386afb3acdf3d927142fd35ac75db344241abc698d4571daac5
against model revision 5763fd50daf633d6203a3b157f85f86d19fa6b38, semantic
scorer sentence-transformers/all-MiniLM-L6-v2, 300 rows per benchmark.

bench-effortless-7-2026 exact_match 0.0200 acc_norm 0.4900
bench-easy-7-2026 hybrid_score 0.3002 acc_norm 0.4733
bench-mid-7-2026 soft_score_norm 0.3867 acc_norm 0.3867

The 6-2026 tiers and bench-AGI are left unevaluated. Only the new entry and the
updated date changed, no existing entry was touched.