Atlas

Benchmarks

← All benchmarks

BIG-Bench Extra Hard

General QA · 2025-02-26

BIG-Bench Extra Hard replaces each BIG-Bench Hard task with a more difficult task probing a similar reasoning ability. Scores report aggregate answer accuracy.

Top models (higher is better)

ModelScore
Gemma 4 31B IT74.4
Gemma 4 26B A4B IT64.8
DiffusionGemma 26B A4B IT59.1
Gemma 4 12B IT53.0
Gemma 4 E4B IT33.1
DeepSeek-V4-Pro-Base29.8
DeepSeek-V4-Flash-Base25.4
Gemma 4 E2B IT21.9
Gemma 3 27B IT19.3
Gemma 3 12B IT16.3
Gemma 2 27B IT14.8
Gemma 3 4B IT11.0
Gemma 2 9B IT9.8
Gemma 3 1B IT7.2
Gemma 2 2B IT5.9
Loading Atlas data…