math benchmark
BIG-Bench Hard Leaderboard
BIG-Bench Hard (BBH) is a subset of 23 challenging BIG-Bench tasks selected because prior language model evaluations did not outperform average human-rater performance. The benchmark contains 6,511 evaluation examples testing various forms of multi-step reasoning including arithmetic, logical reasoning (Boolean expressions, logical deduction), geometric reasoning, temporal reasoning, and language understanding. Tasks require capabilities such as causal judgment, object counting, navigation, pattern recognition, and complex problem solving.
Updated Aug 17, 2026
Models21
Model coverage21
MetricScore
EvidenceB
BIG-Bench Hard Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score93.1% | Percentile100.0% | Participants21 | EvidenceC | Evaluated |
| Rank02 | ModelAN | Score93.1% | Percentile95.0% | Participants21 | EvidenceC | Evaluated |
| Rank03 | ModelGO | Score89.2% | Percentile90.0% | Participants21 | EvidenceC | Evaluated |
| Rank04 | ModelGO | Score87.6% | Percentile85.0% | Participants21 | EvidenceC | Evaluated |
| Rank05 | ModelAN | Score86.8% | Percentile80.0% | Participants21 | EvidenceC | Evaluated |
| Rank06 | ModelGO | Score85.7% | Percentile75.0% | Participants21 | EvidenceC | Evaluated |
| Rank07 | ModelGO | Score85.5% | Percentile70.0% | Participants21 | EvidenceC | Evaluated |
| Rank08 | ModelAN | Score82.9% | Percentile65.0% | Participants21 | EvidenceC | Evaluated |
| Rank09 | ModelMI | Score79.1% | Percentile60.0% | Participants21 | EvidenceC | Evaluated |
| Rank10 | ModelAN | Score73.7% | Percentile55.0% | Participants21 | EvidenceC | Evaluated |
| Rank11 | ModelGO | Score72.2% | Percentile50.0% | Participants21 | EvidenceC | Evaluated |
| Rank12 | ModelMI | Score70.4% | Percentile45.0% | Participants21 | EvidenceC | Evaluated |
| Rank13 | ModelIB | Score69.1% | Percentile40.0% | Participants21 | EvidenceC | Evaluated |
| Rank14 | ModelIB | Score69.1% | Percentile35.0% | Participants21 | EvidenceC | Evaluated |
| Rank15 | ModelMI | Score69.0% | Percentile30.0% | Participants21 | EvidenceC | Evaluated |
| Rank16 | ModelIB | Score55.7% | Percentile25.0% | Participants21 | EvidenceC | Evaluated |
| Rank17 | ModelGO | Score52.9% | Percentile20.0% | Participants21 | EvidenceC | Evaluated |
| Rank18 | ModelGO | Score52.9% | Percentile15.0% | Participants21 | EvidenceC | Evaluated |
| Rank19 | ModelGO | Score44.3% | Percentile10.0% | Participants21 | EvidenceC | Evaluated |
| Rank20 | ModelGO | Score44.3% | Percentile5.0% | Participants21 | EvidenceC | Evaluated |
| Rank21 | ModelGO | Score39.1% | Percentile0.0% | Participants21 | EvidenceC | Evaluated |
BIG-Bench Hard Highlights
The leading models and scores on this benchmark.
BIG-Bench Hard Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for BIG-Bench Hard
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis big-bench hard AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01ANClaude 3.5 SonnetAnthropic
- Score
- 93.1%
- Speed
- Up to 101 tok/s via Bedrock
Strengths
- Ranks #1 of 21 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures BIG-Bench Hard, not total model capability
- 02ANClaude 3.5 SonnetAnthropic
- Score
- 93.1%
- Speed
- Up to 101 tok/s via Bedrock
Strengths
- Ranks #2 of 21 compared models
- 95th percentile on this benchmark
- C evidence result
Considerations
- This result measures BIG-Bench Hard, not total model capability
- 03GOGemini 1.5 ProGoogle
- Score
- 89.2%
- Speed
- Up to 85 tok/s via Google
Strengths
- Ranks #3 of 21 compared models
- 90th percentile on this benchmark
- C evidence result
Considerations
- This result measures BIG-Bench Hard, not total model capability
- 04GOGemma 3 27BGoogle
- Score
- 87.6%
- Speed
- Up to 33 tok/s via DeepInfra
Strengths
- Ranks #4 of 21 compared models
- 85th percentile on this benchmark
- C evidence result
Considerations
- This result measures BIG-Bench Hard, not total model capability
- 05ANClaude 3 OpusAnthropic
- Score
- 86.8%
- Speed
- Up to 120 tok/s via Bedrock
Strengths
- Ranks #5 of 21 compared models
- 80th percentile on this benchmark
- C evidence result
Considerations
- This result measures BIG-Bench Hard, not total model capability
Selection summary
Best AI Models for BIG-Bench Hard
Claude 3.5 Sonnet currently leads BIG-Bench Hard with 93.1%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is BIG-Bench Hard?
What BIG-Bench Hard measures and how its scores work.
BIG-Bench Hard (BBH) is a subset of 23 challenging BIG-Bench tasks selected because prior language model evaluations did not outperform average human-rater performance. The benchmark contains 6,511 evaluation examples testing various forms of multi-step reasoning including arithmetic, logical reasoning (Boolean expressions, logical deduction), geometric reasoning, temporal reasoning, and language understanding. Tasks require capabilities such as causal judgment, object counting, navigation, pattern recognition, and complex problem solving.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- BIG-Bench Hard
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- big-bench-hard|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about BIG-Bench Hard.
Which model scores highest on BIG-Bench Hard?
Claude 3.5 Sonnet is currently ranked first with 93.1%.
What does BIG-Bench Hard measure?
BIG-Bench Hard (BBH) is a subset of 23 challenging BIG-Bench tasks selected because prior language model evaluations did not outperform average human-rater performance. The benchmark contains 6,511 evaluation examples testing various forms of multi-step reasoning including arithmetic, logical reasoning (Boolean expressions, logical deduction), geometric reasoning, temporal reasoning, and language understanding. Tasks require capabilities such as causal judgment, object counting, navigation, pattern recognition, and complex problem solving.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
21 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.