# IFEval leaderboard

Instruction Following benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelchoice.dev/leaderboards/benchmarks/instruction-following-ifeval

Instruction-Following Eval. Every model the catalog carries a published IFEval value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Instruction Following |
| Measure | Constrained generation |
| Tasks | 541 prompts across 25 instruction types |
| Difficulty | Instruction precision |
| Published by | [Instruction-Following Evaluation for Large Language Models](https://arxiv.org/abs/2311.07911) |

A benchmark of 541 prompts built from 25 verifiable instruction types. It tests whether a model follows checkable constraints such as keyword, length, casing, and response-format requirements.

## IFEval ranking

70 models with a published IFEval value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Constrained generation | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [LFM2.5-24B-A2B](https://modelchoice.dev/models/lfm2-5-24b-a2b) | LiquidAI | 96.75 | Observed 2026-11-21 · source benchlm:models |
| 2 | [LFM3-1.7B](https://modelchoice.dev/models/lfm3-1-7b) | LiquidAI | 96.66 | Observed 2026-11-21 · source benchlm:models |
| 3 | [Flint](https://modelchoice.dev/models/flint-40b-a2-5b) | Recursive | 95.1 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Qwen3.5-27B](https://modelchoice.dev/models/qwen3-5-27b) | Alibaba | 95 | Observed 2026-11-21 · source benchlm:models |
| 5 | [Agents-A1](https://modelchoice.dev/models/agents-a1) | InternScience | 94.82 | Observed 2026-11-21 · source benchlm:models |
| 6 | [Agents-A1-4B](https://modelchoice.dev/models/agents-a1-4b) | InternScience | 94.8 | Observed 2026-11-21 · source benchlm:models |
| 7 | [Qwen3.7 Plus](https://modelchoice.dev/models/qwen3-7-plus) | Alibaba | 94.6 | Observed 2026-11-21 · source benchlm:models |
| 8 | [Qwen3.6 Plus](https://modelchoice.dev/models/qwen3-6-plus) | Alibaba | 94.3 | Observed 2026-11-21 · source benchlm:models |
| 8 | [Qwen3.7 Max](https://modelchoice.dev/models/qwen3-7-max) | Alibaba | 94.3 | Observed 2026-11-21 · source benchlm:models |
| 10 | [dots3-note Preview](https://modelchoice.dev/models/dots3-note-preview) | Dots Studio | 93.9 | Observed 2026-11-21 · source benchlm:models |
| 10 | [Kimi K2.5](https://modelchoice.dev/models/kimi-k2-5) | Moonshot AI | 93.9 | Observed 2026-11-21 · source benchlm:models |
| 10 | [o3-mini](https://modelchoice.dev/models/o3-mini) | OpenAI | 93.9 | Observed 2026-11-21 · source benchlm:models |
| 13 | [Spark-X2.6-4B](https://modelchoice.dev/models/spark-x2-6-4b) | SparkLLM | 93.6 | Observed 2026-11-21 · source benchlm:models |
| 14 | [Qwen3.5-122B-A10B](https://modelchoice.dev/models/qwen3-5-122b-a10b) | Alibaba | 93.4 | Observed 2026-11-21 · source benchlm:models |
| 15 | [Claude 3.7 Sonnet Thinking](https://modelchoice.dev/models/claude-3-7-sonnet-thinking) | Anthropic | 93.2 | Observed 2026-11-21 · source benchlm:models |
| 16 | [GLM-5](https://modelchoice.dev/models/glm-5) | Z.AI | 92.6 | Observed 2026-11-21 · source benchlm:models |
| 16 | [Qwen3.5 397B](https://modelchoice.dev/models/qwen3-5-397b) | Alibaba | 92.6 | Observed 2026-11-21 · source benchlm:models |
| 18 | [K-EXAONE 2.0](https://modelchoice.dev/models/k-exaone-2-0) | LG AI Research | 92.4 | Observed 2026-11-21 · source benchlm:models |
| 19 | [o1](https://modelchoice.dev/models/o1) | OpenAI | 92.2 | Observed 2026-11-21 · source benchlm:models |
| 20 | [Llama 3.3 70B](https://modelchoice.dev/models/llama-3-3-70b) | Meta | 92.1 | Observed 2026-11-21 · source benchlm:models |
| 21 | [Qwen3.5-35B-A3B](https://modelchoice.dev/models/qwen3-5-35b-a3b) | Alibaba | 91.9 | Observed 2026-11-21 · source benchlm:models |
| 22 | [LFM2.5-8B-A1B](https://modelchoice.dev/models/lfm2-5-8b-a1b) | LiquidAI | 91.84 | Observed 2026-11-21 · source benchlm:models |
| 23 | [Ternary Bonsai 2 27B](https://modelchoice.dev/models/ternary-bonsai-2-27b) | Prism ML | 91.31 | Observed 2026-11-21 · source benchlm:models |
| 24 | [Claude Opus 4.5](https://modelchoice.dev/models/claude-opus-4-5) | Anthropic | 90.9 | Observed 2026-11-21 · source benchlm:models |
| 25 | [Claude 3.7 Sonnet](https://modelchoice.dev/models/claude-3-7-sonnet) | Anthropic | 90.8 | Observed 2026-11-21 · source benchlm:models |
| 26 | [Mellum2.1-12B-A2.5B-Thinking](https://modelchoice.dev/models/mellum2-1-12b-a2-5b-thinking) | JetBrains | 90.6 | Observed 2026-11-21 · source benchlm:models |
| 27 | [Kimi K2](https://modelchoice.dev/models/kimi-k2) | Moonshot AI | 89.8 | Observed 2026-11-21 · source benchlm:models |
| 28 | [Nemotron Ultra 253B](https://modelchoice.dev/models/nemotron-ultra-253b) | NVIDIA | 89.45 | Observed 2026-11-21 · source benchlm:models |
| 29 | [Qwen3 235B 2507](https://modelchoice.dev/models/qwen3-235b-2507) | Alibaba | 88.7 | Observed 2026-11-21 · source benchlm:models |
| 30 | [Llama 3.1 405B](https://modelchoice.dev/models/llama-3-1-405b) | Meta | 88.6 | Observed 2026-11-21 · source benchlm:models |
| 31 | [GPT-4.1 mini](https://modelchoice.dev/models/gpt-4-1-mini) | OpenAI | 88.5 | Observed 2026-11-21 · source benchlm:models |
| 32 | [Qwen3 235B 2507 (Reasoning)](https://modelchoice.dev/models/qwen3-235b-2507-reasoning) | Alibaba | 87.8 | Last good 2026-11-18 · source benchlm:models |
| 33 | [Llama 3.1 70B](https://modelchoice.dev/models/llama-3-1-70b) | Meta | 87.5 | Observed 2026-11-21 · source benchlm:models |
| 34 | [GPT-4.1](https://modelchoice.dev/models/gpt-4-1) | OpenAI | 87.4 | Observed 2026-11-21 · source benchlm:models |
| 35 | [MiniCPM5-2B](https://modelchoice.dev/models/minicpm5-2b) | OpenBMB | 86.7 | Observed 2026-11-21 · source benchlm:models |
| 36 | [DeepSeek V3](https://modelchoice.dev/models/deepseek-v3) | DeepSeek | 86.1 | Observed 2026-11-21 · source benchlm:models |
| 37 | [Claude 3.5 Haiku](https://modelchoice.dev/models/claude-3-5-haiku) | Anthropic | 85.9 | Observed 2026-11-21 · source benchlm:models |
| 38 | [ZAYA1-8B](https://modelchoice.dev/models/zaya1-8b) | Zyphra | 85.58 | Observed 2026-11-21 · source benchlm:models |
| 39 | [Qwen3 32B (Reasoning)](https://modelchoice.dev/models/qwen3-32b-reasoning) | Alibaba | 85 | Observed 2026-11-21 · source benchlm:models |
| 40 | [Phi-4-reasoning-plus](https://modelchoice.dev/models/phi-4-reasoning-plus) | Microsoft | 84.9 | Observed 2026-11-21 · source benchlm:models |
| 41 | [Sarvam 105B](https://modelchoice.dev/models/sarvam-105b) | Sarvam | 84.8 | Observed 2026-11-21 · source benchlm:models |
| 42 | [QwQ-32B](https://modelchoice.dev/models/qwq-32b) | Alibaba | 83.9 | Observed 2026-11-21 · source benchlm:models |
| 43 | [Phi-4-reasoning](https://modelchoice.dev/models/phi-4-reasoning) | Microsoft | 83.4 | Observed 2026-11-21 · source benchlm:models |
| 43 | [Qwen3 235B-A22B (Reasoning)](https://modelchoice.dev/models/qwen3-235b-a22b-reasoning) | Alibaba | 83.4 | Observed 2026-11-21 · source benchlm:models |
| 45 | [DeepSeek-R1](https://modelchoice.dev/models/deepseek-r1) | DeepSeek | 83.3 | Observed 2026-11-21 · source benchlm:models |
| 46 | [GPT-4.1 nano](https://modelchoice.dev/models/gpt-4-1-nano) | OpenAI | 83.2 | Observed 2026-11-21 · source benchlm:models |
| 46 | [Qwen3 235B-A22B](https://modelchoice.dev/models/qwen3-235b-a22b) | Alibaba | 83.2 | Observed 2026-11-21 · source benchlm:models |
| 46 | [Qwen3 32B](https://modelchoice.dev/models/qwen3-32b) | Alibaba | 83.2 | Observed 2026-11-21 · source benchlm:models |
| 49 | [OrcaCyber-Nano-1.0](https://modelchoice.dev/models/orcacyber-nano-1-0) | OrcaRouter | 82.81 | Observed 2026-11-21 · source benchlm:models |
| 50 | [LFM2.5-VL-3B](https://modelchoice.dev/models/lfm2-5-vl-3b) | LiquidAI | 82.3 | Observed 2026-11-21 · source benchlm:models |
| 51 | [Ternary Bonsai 8B](https://modelchoice.dev/models/ternary-bonsai-8b) | Prism ML | 81.8 | Observed 2026-11-21 · source benchlm:models |
| 52 | [Kanana-2 3B Instruct](https://modelchoice.dev/models/kanana-2-3b-instruct) | Kakao | 80.96 | Observed 2026-11-21 · source benchlm:models |
| 53 | [Celeris-1](https://modelchoice.dev/models/celeris-1) | Celeris | 80.8 | Observed 2026-11-21 · source benchlm:models |
| 54 | [MiniCPM5-1B](https://modelchoice.dev/models/minicpm5-1b) | OpenBMB | 80.41 | Observed 2026-11-21 · source benchlm:models |
| 55 | [Llama 3.1 8B](https://modelchoice.dev/models/llama-3-1-8b) | Meta | 80.4 | Observed 2026-11-21 · source benchlm:models |
| 56 | [1-bit Bonsai 8B](https://modelchoice.dev/models/bonsai-8b) | Prism ML | 79.8 | Observed 2026-11-21 · source benchlm:models |
| 57 | [Granite-4.0-H-1B](https://modelchoice.dev/models/granite-4-0-h-1b) | IBM | 78.53 | Observed 2026-11-21 · source benchlm:models |
| 58 | [Kanana-2 1.3B Instruct](https://modelchoice.dev/models/kanana-2-1-3b-instruct) | Kakao | 77.63 | Observed 2026-11-21 · source benchlm:models |
| 59 | [Granite-4.0-1B](https://modelchoice.dev/models/granite-4-0-1b) | IBM | 77.38 | Observed 2026-11-21 · source benchlm:models |
| 60 | [LFM2.5-350M](https://modelchoice.dev/models/lfm2-5-350m) | LiquidAI | 76.96 | Observed 2026-11-21 · source benchlm:models |
| 61 | [Mellum2-12B-A2.5B-Thinking](https://modelchoice.dev/models/mellum2-12b-a2-5b-thinking) | JetBrains | 76.5 | Observed 2026-11-21 · source benchlm:models |
| 62 | [Mellum2-12B-A2.5B-Instruct](https://modelchoice.dev/models/mellum2-12b-a2-5b-instruct) | JetBrains | 75.8 | Observed 2026-11-21 · source benchlm:models |
| 63 | [Ternary Bonsai 4B](https://modelchoice.dev/models/ternary-bonsai-4b) | Prism ML | 72.1 | Observed 2026-11-21 · source benchlm:models |
| 64 | [LFM2.5-230M](https://modelchoice.dev/models/lfm2-5-230m) | LiquidAI | 71.71 | Observed 2026-11-21 · source benchlm:models |
| 65 | [Ternary Bonsai 1.7B](https://modelchoice.dev/models/ternary-bonsai-1-7b) | Prism ML | 70.1 | Observed 2026-11-21 · source benchlm:models |
| 66 | [1-bit Bonsai 4B](https://modelchoice.dev/models/bonsai-4b) | Prism ML | 69.6 | Observed 2026-11-21 · source benchlm:models |
| 67 | [1-bit Bonsai 1.7B](https://modelchoice.dev/models/bonsai-1-7b) | Prism ML | 63 | Observed 2026-11-21 · source benchlm:models |
| 68 | [Granite-4.0-H-350M](https://modelchoice.dev/models/granite-4-0-h-350m) | IBM | 61.63 | Observed 2026-11-21 · source benchlm:models |
| 69 | [LFM2.5-VL-450M](https://modelchoice.dev/models/lfm2-5-vl-450m) | LiquidAI | 61.16 | Observed 2026-11-21 · source benchlm:models |
| 70 | [Granite-4.0-350M](https://modelchoice.dev/models/granite-4-0-350m) | IBM | 55.4 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
