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

IFBench

74 models

IFBench (Instruction-Following Benchmark) measures an AI's ability to follow complex, multi-constraint instructions precisely — including specific formatting, length restrictions, content requirements, and logical conditions — testing adherence to user-provided rules rather than just factual accuracy.

Top 5 Models Performance

recursivesi: flint ######################################## 89.2
krafton: raon-1-mini-preview ###################################### 84.2
MiniMax: MiniMax M3 ##################################### 83
SpaceXAI: Grok 4.20 ##################################### 83
NVIDIA Corporation: nvidia-nemotron-3-ultra ##################################### 81.7
– 970 – 1.1T
– 2019 – 2026
Rank Model Score
❓ 1 recursivesi: flint 89.2
❓ 2 krafton: raon-1-mini-preview 84.2
🇨🇳 3 MiniMax: MiniMax M3稀宇科技 83
🇺🇸 4 SpaceXAI: Grok 4.20 83
🇺🇸 5 NVIDIA Corporation: nvidia-nemotron-3-ultra 81.7
❓ 6 internscience: sage-a1.5 81.6
🇨🇳 7 Qwen: qwen3.8-flash-next通义千问 81.3
❓ 8 NousResearch: hermes-3-llama-3.1-70b 81.2
🇺🇸 9 SpaceXAI: Grok 4.3 81
🇺🇸 10 Amazon: nova-2.0-pro 80.2
🇨🇳 11 Xiaomi: MiMo-V2.5 Pro小米集团 80
🇨🇳 12 Qwen: qwen3.8-27b通义千问 79.5
❓ 13 recursiveai: ripple-65b-preview 79.2
🇨🇳 14 Qwen: Qwen3.7 Max通义千问 79.1
🇨🇳 15 Qwen: qwen3.7-plus通义千问 79.1
🇨🇳 16 DeepSeek: DeepSeek V4 Flash深度求索 79
🇺🇸 17 Meta: muse-glimmer-30b 77
🇨🇳 18 Qwen: Qwen3.5 27B通义千问 77
🇺🇸 19 Meta: Muse Spark 76
❓ 20 xhtoken: spark-x2.6-4b 75.8
❓ 21 xhtoken: spark-x2.5-4b 75
🇺🇸 22 OpenAI: gpt-5.3-codex 75
🇺🇸 23 LiquidAI: lfm-2.5-24b-a2b 74.69
🇨🇳 24 inclusionAI: ling-3.0-flash 74.5
❓ 25 weiboai: vibethinker-3b 74.5
❓ 26 webai-official: twil-8b 74.4
❓ 27 Prism ML: ternary-bonsai-2-27b 74
🇺🇸 28 LiquidAI: lfm-3-1.7b 73.03
🇨🇳 29 Z.ai: GLM 5.2北京智谱华章科技有限公司 73
❓ 30 diffuspace: clover-35b-a3b-preview 72.3
🇨🇳 31 Z.ai: GLM 5北京智谱华章科技有限公司 72.3
🇺🇸 32 NVIDIA Corporation: nvidia-nemotron-3.5-lightning-30b-a3b-bf16 71.88
🇺🇸 33 NVIDIA Corporation: nvidia-nemotron-3-nano-30b-a3b 71.5
🇨🇳 34 Intern Language Models: intern-s1-pro 71.2
🇺🇸 35 Inception: Mercury 2 71
🇺🇸 36 NVIDIA Corporation: nvidia-nemotron-3.5-super-vl-120b-a12b 70.86
❓ 37 xingchen-agi: xing4.0-29b-a4b 69.67
🇨🇳 38 Xiaomi: mimo-v2-pro小米集团 69
🇫🇷 39 Mistral AI: mistral-medium-3.5-128b 69
🇰🇷 40 LG AI Research: k-exaone-236b-a23b주식회사 엘지 67.3
🇨🇳 41 Xiaomi: MiMo-V2.5小米集团 67
🇨🇳 42 OpenBMB: minicpm5-2b 66.3
❓ 43 xhtoken: spark-x2.5-1.7b 66.3
🇺🇸 44 NVIDIA Corporation: nvidia-nemotron-3.5-bolt-e4b 63.79
🇨🇳 45 inclusionAI: ling-3.0-tiny 63.61
🇺🇸 46 Anthropic: Claude Fable 5 63
🇺🇸 47 Anthropic: Claude Opus 4.8 62
❓ 48 Prism ML: ternary-bonsai-27b 58.5
🇺🇸 49 LiquidAI: lfm-2.5-8b-a1b 56.47
❓ 50 orcarouter: orcacyber-nano-1.0 56.33
🇦🇪 51 Technology Innovation Institute: falcon-h1r-7b 53.4
🇺🇸 52 Google: Gemma 4 31B Instruct 52.3
🇺🇸 53 Arcee AI: trinity-large-thinking 52.3
❓ 54 zyphra: zaya1-8b 52.26
🇺🇸 55 Meta Llama: Llama 3.3 70B Instruct 47
🇺🇸 56 SpaceXAI: Grok 3 47
🇨🇳 57 OpenBMB: minicpm5-1b 46.67
🇺🇸 58 SpaceXAI: Grok 3 Mini 46
🇷🇺 59 Sber: gigachat3.5-432b-a28bСбер 43.66
🇺🇸 60 LiquidAI: lfm-2.5-350m 40.69
❓ 61 Ornith: ornith-1.5-9b 39
🇺🇸 62 LiquidAI: LFM2.5-230M 38.4
🇺🇸 63 NVIDIA Corporation: cosmos3-super 37
🇨🇳 64 Qwen: qwen3-vl-32b-instruct通义千问 37
🇺🇸 65 NVIDIA Corporation: cosmos-reason2-32b 28.2
🇺🇸 66 Hugging Face Smol Models Research: smollm3-3b 27.6
🇨🇳 67 Tencent: hunyuan-4b-instruct腾讯控股有限公司 26.5
🇺🇸 68 LiquidAI: lfm-2-8b-a1b 25.85
🇺🇸 69 Google: Gemma 3 4B Instruct 23.53
🇺🇸 70 Microsoft: phi-4-mini-instruct 21.1
🇨🇳 71 Qwen: Qwen3.5 0.8B通义千问 21
🇺🇸 72 Meta Llama: Llama 3.2 3B Instruct 20.78
🇺🇸 73 Google: Gemma 3 1B Instruct 20.33
🇨🇦 74 Cohere Labs: Tiny Aya Global 20