PIQA leaderboard
As of 6 Nov 2026, the top score on PIQA is 88.7% by GPT-4o mini (OpenAI), ahead of Phi-3.5-MoE at 88.6%, across 56 model configurations with a published result (Epoch AI Benchmarking Hub, result dated 18 Jul 2024). Physical commonsense questions. PIQA is shown for reference only and does not count toward the BenchLeader Index.
Key facts (as of 6 Nov 2026)
- GPT-4o mini holds the top PIQA score at 88.7% (Epoch AI Benchmarking Hub, 18 Jul 2024).
- Phi-3.5-MoE is second on PIQA at 88.6% (Epoch AI Benchmarking Hub, 17 Aug 2024).
- PIQA: Physical commonsense questions. A 2019 benchmark, saturated for current models; kept for history (Epoch AI Benchmarking Hub).
- 56 model configurations have a published PIQA result on BenchLeader (BenchLeader, 6 Nov 2026).
- PIQA does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).
Physical commonsense questions. A 2019 benchmark, saturated for current models; kept for history.
- Published by
- Epoch AI Benchmarking Hub
- Category
- Reasoning
- Index weight
- Reference only
- Models
- 56
- Data as of
- 6 Nov 2026
CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.
What the test looks like
Physical commonsense questions.
How it is scored
Accuracy, as compiled by Epoch AI from published results.
What to keep in mind
Saturated: frontier models score near the ceiling, so it separates only older or smaller models, and results are largely self-reported by developers.
- 1GPT-4o mini88.7%
- 2Phi-3.5-MoE88.6%
- 3Gemini 1.5 Flash 00287.5%
- 4Llama 3.1 405B85.9%
- 5Falcon-180B84.9%
- 6Deepseek v3 Base84.7%
- 7Inflection-184.2%
- 8Deepseek v283.9%
- 9Gemma 2 9B83.7%
- 10Mixtral 8x7B83.6%
- 11Mistral Nemo Base 240783.5%
- 12Stable Beluga 283.3%
- 13Megatron-Turing NLG 530B83.2%
- 14Mistral 7B v0.383.0%
- 15Falcon-40B83.0%
| # | ||||
|---|---|---|---|---|
| 1 |  GPT-4o mini | 88.7% | 35.4 | 2024-07-18 |
| 2 |  Phi-3.5-MoE | 88.6% | – | 2024-08-17 |
| 3 |  Gemini 1.5 Flash 002 | 87.5% | 38.2 | 2024-09-24 |
| 4 |  Llama 3.1 405Bopen ↗ | 85.9% | 41.9 | 2024-07-23 |
| 5 |  Falcon-180B | 84.9% | – | 2023-09-06 |
| 6 |  Deepseek v3 Base | 84.7% | – | 2024-12-26 |
| 7 |  Inflection-1 | 84.2% | – | 2023-06-22 |
| 8 | Deepseek v2open ↗ | 83.9% | – | 2024-05-07 |
| 9 | Gemma 2 9B | 83.7% | 38.8 | – |
| 10 |  Mixtral 8x7Bopen | 83.6% | 30.9 | 2023-12-11 |
| 11 | Mistral Nemo Base 2407 | 83.5% | – | 2024-07-18 |
| 12 |  Stable Beluga 2open | 83.3% | – | 2023-07-20 |
| 13 |  Megatron-Turing NLG 530B | 83.2% | – | 2021-10-11 |
| 14 | Mistral 7B v0.3open | 83.0% | 29.0 | 2023-09-27 |
| 15 | Falcon-40B | 83.0% | – | 2023-03-15 |
| 16 | Llama 2-70Bopen ↗ | 82.8% | 33.6 | 2023-07-18 |
| 17 | LLaMA-65Bopen | 82.8% | – | 2023-02-24 |
| 18 |  Qwen2.5 72Bopen ↗ | 82.6% | 43.2 | 2024-09-19 |
| 19 | Nemotron-4 15B | 82.4% | – | 2024-02-26 |
| 20 | InstructGPT 175B | 82.3% | – | 2022-01-27 |
| 21 | LLaMA-33B | 82.3% | – | 2023-02-24 |
| 22 | PaLM 540B | 82.3% | – | 2022-04-04 |
| 23 |  MPT-30B | 81.9% | – | 2023-06-22 |
| 24 | Llama 2-34B | 81.9% | – | 2023-07-18 |
| 25 | Llama 3.1 8Bopen ↗ | 81.2% | 33.7 | 2024-07-23 |
| 26 | Gemma 7B | 81.2% | – | 2024-02-21 |
| 27 | Phi-3.5-mini | 81.0% | – | 2024-08-16 |
| 28 | Llama 2-13Bopen ↗ | 80.8% | – | 2023-07-18 |
| 29 | MPT-7B | 80.6% | – | 2023-05-05 |
| 30 | PaLM 62B | 80.5% | – | 2022-04-04 |
| 31 | Falcon-7B | 80.3% | – | 2023-04-24 |
| 32 |  InternLM 20B | 80.3% | – | 2023-09-18 |
| 33 | LLaMA-13B | 80.1% | – | 2023-02-24 |
| 34 | Qwen-14B | 79.9% | – | 2023-09-28 |
| 35 | LLaMA-7B | 79.8% | – | 2023-02-24 |
| 36 | Llama 2-7Bopen ↗ | 78.8% | – | 2023-07-18 |
| 37 |  Baichuan2-13Bopen | 78.1% | – | 2023-09-06 |
| 38 | InternLM 7B | 77.9% | – | 2023-07-05 |
| 39 | Qwen-7B | 77.9% | – | 2023-09-28 |
| 40 |  Vicuna 13B v1.1 | 77.4% | – | 2023-04-12 |
| 41 | Gemma 2B | 77.3% | – | 2024-02-21 |
| 42 |  RedPajama-INCITE-7B-Base | 76.9% | – | 2023-05-04 |
| 43 |  GPT-NeoX-20B | 76.7% | – | 2022-04-07 |
| 44 | Baichuan1-7Bopen | 76.2% | – | 2023-06-01 |
| 45 | open_llama_7b | 76.0% | – | 2023-06-07 |
| 46 | OPT-13B | 75.7% | – | 2022-05-11 |
| 47 |  Xgen 7B Base | 75.5% | – | 2023-06-27 |
| 48 |  Dolly v2 12B | 75.4% | – | 2023-04-11 |
| 49 | GPT-J-6B | 75.4% | – | 2021-08-05 |
| 50 |  Cerebras-GPT-13B | 73.5% | – | 2023-03-20 |
| 51 | Qwen-1_8B | 73.3% | – | 2023-11-30 |
| 52 | GPT-Neo-2.7B | 72.9% | – | 2021-03-30 |
| 53 | Gpt2 Xl | 70.5% | – | 2019-11-05 |
| 54 |  ChatGLM2-6B | 69.6% | – | 2023-06-24 |
| 55 | OPT-1.3B | 69.0% | – | 2022-05-11 |
| 56 | StableLM Tuned Alpha 7B | 65.8% | – | 2023-04-19 |
More reasoning benchmarks · All LLM benchmarks
Related benchmarks
Cite as: BenchLeader, “PIQA leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/piqa.
PIQA: questions
- What does PIQA measure?
- Physical commonsense questions. Scores are reported in percent of tasks solved; higher is better.
- What is the top score on PIQA?
- As of 6 Nov 2026, the highest score on PIQA is 88.7% by GPT-4o mini (result dated 18 Jul 2024), ahead of Phi-3.5-MoE at 88.6%. Scores come from Epoch AI Benchmarking Hub and are re-read daily.
- Which AI model leads PIQA?
- GPT-4o mini leads PIQA with 88.7% as of 6 Nov 2026 (result dated 18 Jul 2024), ahead of Phi-3.5-MoE at 88.6%.
- How many models have PIQA results?
- 56 model configurations have a PIQA result on BenchLeader, all taken from Epoch AI Benchmarking Hub.
- Who runs PIQA and how often is it updated?
- PIQA is published by Epoch AI Benchmarking Hub. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does PIQA count toward the BenchLeader Index?
- No. PIQA is shown for reference but left out of the composite index.