Benchgen

BIG-Bench Hard — Results

RankModelScore
1ling-3-0-flash-base89.17
2deepseek-v4-1-flash86.1
3ling-3-0-tiny-base80.64
4claude-3-5-sonnet0.931
5gemini-1-5-pro0.892
6claude-3-opus0.868
7gemini-1-5-flash0.855
8phi-3-5-moe-instruct0.791
9phi-4-mini0.704
10phi-3-5-mini-instruct0.69

BIG-Bench Hard — Phases

1 phaseActive

Phase 1·

BIG-Bench Hard

Aug 3, 2026
Current