← Benchmarks
IFEval
Instruction-Following Evaluation (IFEval) benchmark for large language models, focusing on verifiable instructions with 25 types of instructions and around 500 prompts containing one or more verifiable constraints
id ifeval · max 1 · 70 models reported
| # | Model | Score |
|---|---|---|
| 1 | Qwen3.5-27B Alibaba Cloud / Qwen Team · open | 0.95 |
| 2 | Qwen3.7-Plus Alibaba Cloud / Qwen Team | 0.95 |
| 3 | Qwen3.6 Plus Alibaba Cloud / Qwen Team | 0.94 |
| 4 | Qwen3.7 Max Alibaba Cloud / Qwen Team | 0.94 |
| 5 | o3-mini OpenAI | 0.94 |
| 6 | Qwen3 Max Thinking Alibaba Cloud / Qwen Team | 0.93 |
| 7 | Qwen3.5-122B-A10B Alibaba Cloud / Qwen Team · open | 0.93 |
| 8 | Claude 3.7 Sonnet Anthropic | 0.93 |
| 9 | Qwen3.5-397B-A17B Alibaba Cloud / Qwen Team · open | 0.93 |
| 10 | Llama 3.3 70B Instruct Meta · open | 0.92 |
| 11 | Nova Pro Amazon | 0.92 |
| 12 | Qwen3.5-35B-A3B Alibaba Cloud / Qwen Team · open | 0.92 |
| 13 | Qwen3.5-9B Alibaba Cloud / Qwen Team · open | 0.92 |
| 14 | Gemma 3 27B Google · open | 0.90 |
| 15 | Nemotron Nano 9B v2 NVIDIA · open | 0.90 |
| 16 | Gemma 3 4B Google · open | 0.90 |
| 17 | Kimi K2 Instruct Moonshot AI · open | 0.90 |
| 18 | Kimi K2-Instruct-0905 Moonshot AI · open | 0.90 |
| 19 | Qwen3.5-4B Alibaba Cloud / Qwen Team · open | 0.90 |
| 20 | Nova Lite Amazon | 0.90 |
| 21 | LongCat-Flash-Chat Meituan · open | 0.90 |
| 22 | EXAONE 4.5 33B LG AI Research | 0.90 |
| 23 | Llama 3.1 Nemotron Ultra 253B v1 NVIDIA · open | 0.89 |
| 24 | Gemma 3 12B Google · open | 0.89 |
| 25 | Qwen3-Next-80B-A3B-Thinking Alibaba Cloud / Qwen Team · open | 0.89 |
| 26 | Qwen3-235B-A22B-Instruct-2507 Alibaba Cloud / Qwen Team · open | 0.89 |
| 27 | Llama 3.1 405B Instruct Meta · open | 0.89 |
| 28 | GPT-4.5 OpenAI | 0.88 |
| 29 | Qwen3 VL 235B A22B Thinking Alibaba Cloud / Qwen Team · open | 0.88 |
| 30 | Qwen3 VL 235B A22B Instruct Alibaba Cloud / Qwen Team · open | 0.88 |
| 31 | Qwen3 VL 32B Thinking Alibaba Cloud / Qwen Team · open | 0.88 |
| 32 | Qwen3-235B-A22B-Thinking-2507 Alibaba Cloud / Qwen Team · open | 0.88 |
| 33 | Qwen3-Next-80B-A3B-Instruct Alibaba Cloud / Qwen Team · open | 0.88 |
| 34 | Llama 3.1 70B Instruct Meta · open | 0.88 |
| 35 | GPT-4.1 OpenAI | 0.87 |
| 36 | Kimi-k1.5 Moonshot AI | 0.87 |
| 37 | Nova Micro Amazon | 0.87 |
| 38 | DeepSeek-V3 DeepSeek · open | 0.86 |
| 39 | Qwen3 VL 30B A3B Instruct Alibaba Cloud / Qwen Team · open | 0.86 |
| 40 | Qwen3 14B Alibaba Cloud / Qwen Team · open | 0.85 |
| 41 | Phi 4 Reasoning Plus Microsoft · open | 0.85 |
| 42 | Sarvam-105B Sarvam AI · open | 0.85 |
| 43 | Qwen3 VL 32B Instruct Alibaba Cloud / Qwen Team · open | 0.85 |
| 44 | GPT-4.1 mini OpenAI | 0.84 |
| 45 | Qwen2.5 72B Instruct Alibaba Cloud / Qwen Team · open | 0.84 |
| 46 | QwQ-32B Alibaba Cloud / Qwen Team · open | 0.84 |
| 47 | Qwen3 VL 8B Instruct Alibaba Cloud / Qwen Team · open | 0.84 |
| 48 | Phi 4 Reasoning Microsoft · open | 0.83 |
| 49 | Qwen3 VL 8B Thinking Alibaba Cloud / Qwen Team · open | 0.83 |
| 50 | Mistral Small 3 24B Instruct Mistral AI · open | 0.83 |
| 51 | Qwen3 VL 4B Thinking Alibaba Cloud / Qwen Team · open | 0.83 |
| 52 | LFM2.5-VL-3B Liquid AI · open | 0.82 |
| 53 | Qwen3 VL 4B Instruct Alibaba Cloud / Qwen Team · open | 0.82 |
| 54 | Qwen3 VL 30B A3B Thinking Alibaba Cloud / Qwen Team · open | 0.82 |
| 55 | GPT-4o OpenAI | 0.81 |
| 56 | Llama 3.1 8B Instruct Meta · open | 0.80 |
| 57 | Gemma 3 1B Google · open | 0.80 |
| 58 | Llama 3.1 Nemotron Nano 8B V1 NVIDIA · open | 0.79 |
| 59 | Qwen3.5-2B Alibaba Cloud / Qwen Team · open | 0.79 |
| 60 | Llama 3.2 3B Instruct Meta · open | 0.77 |
| 61 | MiniCPM-SALA OpenBMB · open | 0.76 |
| 62 | North Micro Vision Instruct Cohere · open | 0.75 |
| 63 | Granite 3.3 8B Base IBM · open | 0.75 |
| 64 | Granite 3.3 8B Instruct IBM · open | 0.75 |
| 65 | GPT-4.1 nano OpenAI | 0.74 |
| 66 | Qwen2.5 7B Instruct Alibaba Cloud / Qwen Team · open | 0.71 |
| 67 | IBM Granite 4.0 Tiny Preview IBM · open | 0.63 |
| 68 | Phi 4 Microsoft · open | 0.63 |
| 69 | Pixtral-12B Mistral AI · open | 0.61 |
| 70 | Qwen3.5-0.8B Alibaba Cloud / Qwen Team · open | 0.44 |