Reasoning benchmark # BIG-Bench Hard leaderboard > BIG-Bench Hard results for 27 AI models, led by Gemini 1.5 Pro (May 2024) at 89.2%. What the benchmark measures, who runs it, and a source for every score. - Canonical page: https://noometry.com/benchmarks/bbh - Last updated: 2026-10-10 - Title: BIG-Bench Hard Leaderboard (October 2026): Scores by Model As of October 2026, Gemini 1.5 Pro (May 2024) has the highest published BIG-Bench Hard score on Noometry at 89.2%, out of 27 models with results. Last verified October 10, 2026 ## About BIG-Bench Hard 23 challenging BIG-Bench tasks where early models fell short of human raters. - **Category:** [Reasoning](https://noometry.com/best/reasoning) - **Introduced:** 2022 - **Format:** Mixed - **Unit:** Percent (random guessing ≈ 25%) - **Official site:** [github.com](https://github.com/suzgunmirac/BIG-Bench-Hard) ## Top 15 models Top models on BIG-Bench Hard 1. Gemini 1.5 Pro (May 2024) 89.2% 2. DeepSeek-V3 87.5% 3. Llama 3.1-405B 82.9% 4. phi-3-medium 14B 81.4% 5. Qwen2.5 72B Instruct 79.8% 6. Phi 3 Small 8k Instruct 79.1% 7. DeepSeek-V2 (MoE-236B, May 2024) 78.8% 8. GPT-4 75.1% 9. Phi 3 Mini 4k Instruct 71.7% 10. Yi-34B 71.7% 11. Llama 2-70B 64.9% 12. GPT-3.5-turbo 61.6% 13. Phi-2 59.4% 14. Nemotron-4 15B 58.7% 15. Llama 2-13B 58.2% 16. 5060708090 Sponsored placements are available on pages like this one. [Advertise on Noometry](https://noometry.com/advertise) ## All results BIG-Bench Hard results by model | # | Model | Provider | Score | Setting | Source | Date | | --- | --- | --- | --- | --- | --- | --- | | 1 | [Gemini 1.5 Pro (May 2024)](https://noometry.com/models/gemini-1-5-pro) | [![](/logos/google.svg) Google](https://noometry.com/providers/google) | 89.2% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 2 | [DeepSeek-V3](https://noometry.com/models/deepseek-v3) | [![](/logos/deepseek.svg) DeepSeek](https://noometry.com/providers/deepseek) | 87.5% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 3 | [Llama 3.1-405B](https://noometry.com/models/llama-3-1-405b) | [![](/logos/meta.svg) Meta](https://noometry.com/providers/meta) | 82.9% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 4 | [phi-3-medium 14B](https://noometry.com/models/phi-3-medium-14b) | [![](/logos/microsoft.svg) Microsoft](https://noometry.com/providers/microsoft) | 81.4% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 5 | [Qwen2.5 72B Instruct](https://noometry.com/models/qwen2-5-72b-instruct) | [![](/logos/alibaba.svg) Alibaba (Qwen)](https://noometry.com/providers/alibaba) | 79.8% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 6 | [Phi 3 Small 8k Instruct](https://noometry.com/models/phi-3-small-8k-instruct) | [![](/logos/microsoft.svg) Microsoft](https://noometry.com/providers/microsoft) | 79.1% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 7 | [DeepSeek-V2 (MoE-236B, May 2024)](https://noometry.com/models/deepseek-v2) | [![](/logos/deepseek.svg) DeepSeek](https://noometry.com/providers/deepseek) | 78.8% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 8 | [GPT-4](https://noometry.com/models/gpt-4) | [OpenAI](https://noometry.com/providers/openai) | 75.1% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 9 | [Phi 3 Mini 4k Instruct](https://noometry.com/models/phi-3-mini-4k-instruct) | [![](/logos/microsoft.svg) Microsoft](https://noometry.com/providers/microsoft) | 71.7% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 10 | [Yi-34B](https://noometry.com/models/yi-34b) | [01.AI](https://noometry.com/providers/01-ai) | 71.7% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 11 | [Llama 2-70B](https://noometry.com/models/llama-2-70b) | [![](/logos/meta.svg) Meta](https://noometry.com/providers/meta) | 64.9% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 12 | [GPT-3.5-turbo](https://noometry.com/models/gpt-3-5-turbo) | [OpenAI](https://noometry.com/providers/openai) | 61.6% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 13 | [Phi-2](https://noometry.com/models/phi-2) | [![](/logos/microsoft.svg) Microsoft](https://noometry.com/providers/microsoft) | 59.4% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 14 | [Nemotron-4 15B](https://noometry.com/models/nemotron-4-15b) | [![](/logos/nvidia.svg) NVIDIA](https://noometry.com/providers/nvidia) | 58.7% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 15 | [Llama 2-13B](https://noometry.com/models/llama-2-13b) | [![](/logos/meta.svg) Meta](https://noometry.com/providers/meta) | 58.2% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 16 | [Mistral 7B](https://noometry.com/models/mistral-7b) | [![](/logos/mistral.svg) Mistral AI](https://noometry.com/providers/mistral) | 56.1% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 17 | [Gemma 7B](https://noometry.com/models/gemma-7b) | [![](/logos/google.svg) Google](https://noometry.com/providers/google) | 55.1% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 18 | [Qwen-14B](https://noometry.com/models/qwen-14b) | [![](/logos/alibaba.svg) Alibaba (Qwen)](https://noometry.com/providers/alibaba) | 55% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 19 | [Yi 6B](https://noometry.com/models/yi-6b) | [01.AI](https://noometry.com/providers/01-ai) | 47.2% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 20 | [Qwen-7B](https://noometry.com/models/qwen-7b) | [![](/logos/alibaba.svg) Alibaba (Qwen)](https://noometry.com/providers/alibaba) | 45% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 21 | [Llama 2-34B](https://noometry.com/models/llama-2-34b) | [![](/logos/meta.svg) Meta](https://noometry.com/providers/meta) | 44.1% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 22 | [Llama 2-7B](https://noometry.com/models/llama-2-7b) | [![](/logos/meta.svg) Meta](https://noometry.com/providers/meta) | 39.2% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 23 | [Llama 13b](https://noometry.com/models/llama-13b) | [![](/logos/meta.svg) Meta](https://noometry.com/providers/meta) | 37.9% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 24 | [Falcon-40B](https://noometry.com/models/falcon-40b) | [![](/logos/tii.svg) Technology Innovation Institute](https://noometry.com/providers/tii) | 37.1% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 25 | [Gemma 2B](https://noometry.com/models/gemma-2b) | [![](/logos/google.svg) Google](https://noometry.com/providers/google) | 35.2% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 26 | [INTELLECT-1](https://noometry.com/models/intellect-1) | [![](/logos/huggingface.svg) Hugging Face](https://noometry.com/providers/huggingface) | 34.9% | | [Epoch AI](https://epoch.ai/benchmarks) | | | 27 | [Falcon-7B](https://noometry.com/models/falcon-7b) | [![](/logos/tii.svg) Technology Innovation Institute](https://noometry.com/providers/tii) | 28.8% | | [Epoch AI](https://epoch.ai/benchmarks) | | ## Compare the leaders - [Gemini 1.5 Pro (May 2024) vs DeepSeek-V3](https://noometry.com/compare/deepseek-v3-vs-gemini-1-5-pro) - [Gemini 1.5 Pro (May 2024) vs Llama 3.1-405B](https://noometry.com/compare/gemini-1-5-pro-vs-llama-3-1-405b) - [Gemini 1.5 Pro (May 2024) vs phi-3-medium 14B](https://noometry.com/compare/gemini-1-5-pro-vs-phi-3-medium-14b) - [Gemini 1.5 Pro (May 2024) vs Qwen2.5 72B Instruct](https://noometry.com/compare/gemini-1-5-pro-vs-qwen2-5-72b-instruct) - [DeepSeek-V3 vs Llama 3.1-405B](https://noometry.com/compare/deepseek-v3-vs-llama-3-1-405b) - [DeepSeek-V3 vs phi-3-medium 14B](https://noometry.com/compare/deepseek-v3-vs-phi-3-medium-14b) ## Other reasoning benchmarks - [ARC-AGI-2](https://noometry.com/benchmarks/arc-agi-2) - [SimpleBench](https://noometry.com/benchmarks/simplebench) - [Kagi LLM Benchmark](https://noometry.com/benchmarks/kagi-reasoning) - [NYT Connections (extended)](https://noometry.com/benchmarks/nyt-connections) - [ARC-AGI-1](https://noometry.com/benchmarks/arc-agi-1) - [CritPt](https://noometry.com/benchmarks/critpt) - [Chess Puzzles](https://noometry.com/benchmarks/chess-puzzles) - [EnigmaEval](https://noometry.com/benchmarks/enigmaeval) - [Thematic Generalization](https://noometry.com/benchmarks/thematic-generalization) - [LMArena Hard Prompts](https://noometry.com/benchmarks/arena-hard-prompts) - [EBR-Bench](https://noometry.com/benchmarks/ebr-bench) - [LiveBench Reasoning](https://noometry.com/benchmarks/livebench-reasoning) ## Frequently asked questions ### What does BIG-Bench Hard measure? 23 challenging BIG-Bench tasks where early models fell short of human raters. ### Which model has the highest BIG-Bench Hard score? As of October 2026, Gemini 1.5 Pro (May 2024) has the highest published BIG-Bench Hard score on Noometry at 89.2%, out of 27 models with results. ### What is the best open-weight model on BIG-Bench Hard? DeepSeek-V3 has the highest BIG-Bench Hard accuracy among open-weight models at 87.5%, ranking 2 of 27 overall. ### Cite this page Noometry. (2026). BIG-Bench Hard leaderboard. Retrieved October 10, 2026, from https://noometry.com/benchmarks/bbh Quote Noometry with a link back to this page. It is also available in [Markdown](https://noometry.com/md/benchmarks/bbh.md).