language benchmark
BIG-Bench Extra Hard Leaderboard
BIG-Bench Extra Hard (BBEH) is a 23-task benchmark designed to evaluate challenging reasoning abilities across many-hop, causal, spatial, temporal, geometric, linguistic, logical, and humor-understanding tasks.
Updated Sep 5, 2026
Models11
Model coverage11
MetricScore
EvidenceB
BIG-Bench Extra Hard Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelGO | Score74.40% | Percentile100.00% | Participants11 | EvidenceC | Evaluated |
| Rank02 | ModelGO | Score64.80% | Percentile90.00% | Participants11 | EvidenceC | Evaluated |
| Rank03 | ModelGO | Score53.00% | Percentile80.00% | Participants11 | EvidenceC | Evaluated |
| Rank04 | ModelGO | Score47.60% | Percentile70.00% | Participants11 | EvidenceC | Evaluated |
| Rank05 | ModelGO | Score33.10% | Percentile60.00% | Participants11 | EvidenceC | Evaluated |
| Rank06 | ModelGO | Score21.90% | Percentile50.00% | Participants11 | EvidenceC | Evaluated |
| Rank07 | ModelGO | Score19.30% | Percentile40.00% | Participants11 | EvidenceC | Evaluated |
| Rank08 | ModelGO | Score16.30% | Percentile30.00% | Participants11 | EvidenceC | Evaluated |
| Rank09 | ModelGO | Score15.00% | Percentile20.00% | Participants11 | EvidenceC | Evaluated |
| Rank10 | ModelGO | Score11.00% | Percentile10.00% | Participants11 | EvidenceC | Evaluated |
| Rank11 | ModelGO | Score7.20% | Percentile0.00% | Participants11 | EvidenceC | Evaluated |
BIG-Bench Extra Hard Highlights
The leading models and scores on this benchmark.
BIG-Bench Extra Hard Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for BIG-Bench Extra Hard
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is BIG-Bench Extra Hard?
What BIG-Bench Extra Hard measures and how its scores work.
BBEH is a benchmark in the language category that replaces each task in BIG-Bench Hard with a novel, significantly more difficult task probing similar reasoning capabilities.
It measures performance on diverse reasoning skills, including many-hop reasoning, causal understanding, spatial reasoning, temporal arithmetic, geometric reasoning, linguistic reasoning, logic puzzles, and humor understanding, using Score reported as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- BIG-Bench Extra Hard
- Modality
- text
- Primary category
- language
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about BIG-Bench Extra Hard.
Which model scores highest on BIG-Bench Extra Hard?
Gemma 4 31B is currently ranked first with 74.40%.
What are the top three models on BIG-Bench Extra Hard?
The current leaders are Gemma 4 31B (74.40%), Gemma 4 26B-A4B (64.80%), and Gemma 4 12B (53.00%).
Which BIG-Bench Extra Hard model has the lowest official input price?
No matched official input price is currently available.
Which models are fastest among BIG-Bench Extra Hard results?
The fastest matched records are Gemma 3 27B (33.00 tok/s via DeepInfra), Gemma 3 12B (33.00 tok/s via DeepInfra), and Gemma 3 4B (33.00 tok/s via DeepInfra).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.