Note: Overall leaderboard rankings may not reflect true model quality — individual benchmarks give a clearer picture. ARC-Challenge MMLU GPQA GSM8K Artificial Analysis Intelligence Index v4.1
← Back to leaderboard

ARC-Challenge

92 models

ARC-Challenge (AI2 Reasoning Challenge – Challenge Set) measures an AI's ability to answer grade-school science questions that require multi-step reasoning, common sense, and the application of scientific knowledge beyond simple retrieval.

Top 5 Models Performance

Moonshot AI: Kimi K2 Instruct ######################################## 97.3
Xiaomi: MiMo-V2.5 Pro ######################################## 97.2
Meta Llama: Llama 3.1 405B Instruct ######################################## 96.9
Tencent: hy3-preview-base ####################################### 95.99
OpenAI: GPT-3.5 ################################### 85.2
– 970 – 1.1T
– 2019 – 2026
Rank Model Score
🇨🇳 1 Moonshot AI: Kimi K2 Instruct月之暗面 97.3
🇨🇳 2 Xiaomi: MiMo-V2.5 Pro小米集团 97.2
🇺🇸 3 Meta Llama: Llama 3.1 405B Instruct 96.9
🇨🇳 4 Tencent: hy3-preview-base腾讯控股有限公司 95.99
🇺🇸 5 OpenAI: GPT-3.5 85.2
🇨🇳 6 Qwen: qwen-14b通义千问 84.4
🇺🇸 7 Arcee AI: trinity-nano-base 84
❓ 8 sapientinc: hrm-text-1b 81.9
🇨🇳 9 Qwen: qwen3.5-2b-base通义千问 81.2
🇷🇺 10 Yandex: YandexGPT Lite 5Яндекс 80.8
🇨🇳 11 Qwen: qwen-7b通义千问 75.3
🇨🇳 12 Qwen: qwen2.5-1.5b通义千问 70.7
🇺🇸 13 Google: Gemma 3 27B Pretrained 70.6
🇷🇺 14 Sber: gigachat3.5-432b-a28b-baseСбер 70.39
🇺🇸 15 Meta Llama: Llama 3.2 3B 69.1
🇺🇸 16 Google: gemma-3-12b-pt 68.9
🇺🇸 17 Google: Gemma 3n E4B Instruct 61.6
🇺🇸 18 Hugging Face Smol Models Research: smollm2-1.7b-instruct 60.5
🇩🇪 19 OpenGPT-X: Teuken 7B-instruct-v0.6 59
🇨🇳 20 Qwen: qwen3-1.7b-base通义千问 58.9
🇺🇸 21 Amazon: MistralLite 58.874
🇦🇪 22 Technology Innovation Institute: falcon3-mamba-7b-base 56.7
🇺🇸 23 Google: gemma-3-4b-pt 56.2
🇦🇪 24 Technology Innovation Institute: falcon-h1-1.5b-deep-base 55.89
🇩🇪 25 OpenGPT-X: Teuken 7B-base-v0.6 55.8
🇺🇸 26 Microsoft: phi-4-mini-instruct 54.1806
🇨🇳 27 Qwen: qwen3.5-9b-base通义千问 53.8462
🇨🇳 28 Qwen: qwen3-0.6b-base通义千问 53.8
🇦🇪 29 Technology Innovation Institute: falcon-h1-1.5b-base 53.24
🇨🇳 30 Qwen: qwen-1.8b通义千问 53.2
🇺🇸 31 Hugging Face Smol Models Research: SmolLM2 360M Instruct 53
🇺🇸 32 IBM Granite: granite-4.1-3b 52.5
🇺🇸 33 Google: Gemma 3n E2B Instruct 51.7
🇨🇳 34 Qwen: Qwen3.5 9B通义千问 50.5017
🇺🇸 35 Hugging Face Smol Models Research: smollm-360m-instruct 50.1
❓ 36 empero-ai: qwythos-9b-claude-mythos-5-1m 49
🇨🇳 37 Qwen: Qwen3.5 2B通义千问 46.1538
🇺🇸 38 Hugging Face Smol Models Research: smollm3-3b-base 45.8194
🇨🇳 39 Qwen: qwen2.5-0.5b通义千问 45.3
🇷🇺 40 T-Tech: t-lite-it-1.0 45.2
🇨🇳 41 Qwen: Qwen3.5 4B通义千问 45.1505
🇨🇳 42 Qwen: Qwen3 0.6B通义千问 44.3
🇺🇸 43 Hugging Face Smol Models Research: smollm2-135m 43.9
🇦🇪 44 Technology Innovation Institute: falcon-h1-1.5b-deep-instruct 43.86
🇨🇳 45 Qwen: qwen2.5-0.5b-instruct通义千问 43.3
🇦🇪 46 Technology Innovation Institute: falcon-h1-1.5b-instruct 42.06
🇺🇸 47 Hugging Face Smol Models Research: smollm3-3b 41.806
🇺🇸 48 LiquidAI: lfm-2.5-8b-a1b 41.4716
🇷🇺 49 MWS AI: cotype-nano 40.8
🇨🇳 50 Baidu: ernie-4.5-300b-a47b百度 40.6
🇺🇸 51 Meta Llama: Llama 3.2 3B Instruct 40.4682
🇺🇸 52 Apple: openelm-3b-instruct 39.42
🇨🇳 53 DeepSeek: DeepSeek R1 Distill Llama 8B深度求索 39.1304
🇺🇸 54 Google: gemma-3-1b-pt 38.4
🇨🇳 55 Qwen: qwen2-1.5b-instruct通义千问 38.1
🇺🇸 56 Apple: openelm-1_1b-instruct 37.97
🇺🇸 57 Hugging Face Smol Models Research: smollm2-135m-instruct 37.3
🇺🇸 58 LiquidAI: LFM2.5 1.2B Instruct 36.1204
🇩🇴 59 appvoid: arco-3 36.01
🇺🇸 60 Apple: openelm-3b 35.58
🇺🇸 61 AMD: amd-olmo-1b 33.7
🇫🇷 62 Mistral AI: devstral-small-2505 33.1104
🇺🇸 63 LiquidAI: LFM2.5-230M 33.1
🇨🇳 64 Qwen: Qwen3.5 0.8B通义千问 32.4415
🇺🇸 65 Apple: openelm-1_1b 32.34
🇷🇺 66 Sapbot: Grok 4 Distill SmolLM2 135M 32.107
🇺🇸 67 Apple: openelm-270m-instruct 30.55
🇨🇳 68 OpenBMB: minicpm5-1b 30.4348
🇺🇸 69 Apple: openelm-450m-instruct 30.38
❓ 70 Doge Face: Doge 20M 29.8
🇺🇸 71 LiquidAI: lfm-2.5-350m 28.0936
🇺🇸 72 Apple: openelm-450m 27.56
🇦🇪 73 Technology Innovation Institute: Falcon H1 Tiny 90M Instruct 26.8
❓ 74 SupraLabs: Supra-50M-Reasoning 26.54
🇺🇸 75 Apple: openelm-270m 26.45
❓ 76 pleias: baguettotron 26.087
🇨🇳 77 DeepSeek: deepseek-coder-1.3b-instruct深度求索 25.5
🇦🇪 78 Technology Innovation Institute: falcon-h1-tiny-coder-90m 25.1
- 79 NON AI - НЕ ИИ: Random baseline - Случайный счёт 25
🇺🇸 80 Google: Gemma 3 270M Instruct 24.0803
🇺🇸 81 Google: functiongemma-270m-it 24.0803
🇺🇸 82 NVIDIA Corporation: locateanything-3b 23.4
❓ 83 SupraLabs: Supra-1.5 Instruct EXP 22.9
❓ 84 SupraLabs: Supra-50M Instruct 22.87
🇺🇸 85 OpenAI Community: GPT-2 22.7425
🇫🇷 86 BigScience: bloom-560m 21.7391
❓ 87 SupraLabs: Supra-50M BASE 21.59
🇷🇺 88 Sapbot: ToyLlama-50M 20.0669
❓ 89 raincandy-u: rain-100m 19.398
❓ 90 Amir: tiny-llm 19.398
🇺🇸 91 AMD: AMD-135m 19.11
❓ 92 SupraLabs: Supra Mini v6 1M 16.98