# IFBench 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-if-bench

Instruction Following Benchmark. Every model the catalog carries a published IFBench value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Instruction Following |
| Measure | — |
| Tasks | — |
| Difficulty | — |
| Published by | [BenchLM](https://benchlm.ai/benchmarks/ifbench) |

IFBench evaluates precise instruction-following generalization on 58 challenging, verifiable out-of-domain constraints. Unlike IFEval which tests familiar constraint types, IFBench specifically measures how well models follow novel instructions they haven't been optimized for, exposing overfitting to common instruction patterns.

## IFBench ranking

57 models with a published IFBench 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 | IFBench | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Flint](https://modelchoice.dev/models/flint-40b-a2-5b) | Recursive | 89.2 | Observed 2026-11-21 · source benchlm:models |
| 2 | [MAI-Thinking-1](https://modelchoice.dev/models/mai-thinking-1) | Microsoft | 85 | Observed 2026-11-21 · source benchlm:models |
| 3 | [Qwen3.8 Max](https://modelchoice.dev/models/qwen3-8-max) | Alibaba | 82.8 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Inkling-Small](https://modelchoice.dev/models/inkling-small) | Thinking Machines Lab | 82.2 | Observed 2026-11-21 · source benchlm:models |
| 5 | [Nemotron 3 Ultra](https://modelchoice.dev/models/nemotron-3-ultra-500b) | NVIDIA | 81.7 | Observed 2026-11-21 · source benchlm:models |
| 6 | [Sage-A1.5](https://modelchoice.dev/models/sage-a1-5) | InternScience | 81.6 | Observed 2026-11-21 · source benchlm:models |
| 7 | [Qwen3.8-Omni-Flash](https://modelchoice.dev/models/qwen3-8-omni-flash) | Alibaba | 81.5 | Observed 2026-11-21 · source benchlm:models |
| 8 | [Grok 4.3](https://modelchoice.dev/models/grok-4-3) | xAI | 81.3 | Observed 2026-11-21 · source benchlm:models |
| 8 | [Qwen3.8-Flash-Next](https://modelchoice.dev/models/qwen3-8-flash-next) | Alibaba | 81.3 | Observed 2026-11-21 · source benchlm:models |
| 10 | [Beam 1.0 501B-A23B FP8](https://modelchoice.dev/models/reflection-beam-fp8) | Reflection AI | 80.7 | Observed 2026-11-21 · source benchlm:models |
| 11 | [dots3-note Preview](https://modelchoice.dev/models/dots3-note-preview) | Dots Studio | 80.4 | Observed 2026-11-21 · source benchlm:models |
| 12 | [Solar Open 2](https://modelchoice.dev/models/solar-open-2) | Upstage | 80 | Observed 2026-11-21 · source benchlm:models |
| 13 | [Inkling](https://modelchoice.dev/models/inkling) | Thinking Machines Lab | 79.8 | Observed 2026-11-21 · source benchlm:models |
| 14 | [Beam 1.0 501B-A23B](https://modelchoice.dev/models/reflection-beam) | Reflection AI | 79.6 | Observed 2026-11-21 · source benchlm:models |
| 15 | [Qwen3.8-27B](https://modelchoice.dev/models/qwen3-8-27b) | Alibaba | 79.5 | Observed 2026-11-21 · source benchlm:models |
| 16 | [Granite 4.2 8B](https://modelchoice.dev/models/granite-4-2-8b) | IBM | 79.33 | Observed 2026-11-21 · source benchlm:models |
| 17 | [Beam 1.0 501B-A23B NVFP4](https://modelchoice.dev/models/reflection-beam-nvfp4) | Reflection AI | 79.2 | Observed 2026-11-21 · source benchlm:models |
| 17 | [Ripple-65B Preview](https://modelchoice.dev/models/ripple-65b-preview) | Ricursive Intelligence | 79.2 | Observed 2026-11-21 · source benchlm:models |
| 19 | [Qwen3.7 Max](https://modelchoice.dev/models/qwen3-7-max) | Alibaba | 79.1 | Observed 2026-11-21 · source benchlm:models |
| 19 | [Qwen3.7 Plus](https://modelchoice.dev/models/qwen3-7-plus) | Alibaba | 79.1 | Observed 2026-11-21 · source benchlm:models |
| 21 | [Granite 4.2 30B](https://modelchoice.dev/models/granite-4-2-30b) | IBM | 77.17 | Observed 2026-11-21 · source benchlm:models |
| 22 | [Mercury 2.5](https://modelchoice.dev/models/mercury-2-5) | Inception | 77 | Observed 2026-11-21 · source benchlm:models |
| 22 | [Muse Glimmer 30B](https://modelchoice.dev/models/muse-glimmer-30b) | Meta | 77 | Observed 2026-11-21 · source benchlm:models |
| 24 | [Mercury 3](https://modelchoice.dev/models/mercury-3) | Inception | 76.84 | Observed 2026-11-21 · source benchlm:models |
| 25 | [Qwen3.5 397B](https://modelchoice.dev/models/qwen3-5-397b) | Alibaba | 76.5 | Observed 2026-11-21 · source benchlm:models |
| 26 | [Gemini 3.5 Flash](https://modelchoice.dev/models/gemini-3-5-flash) | Google | 76.3 | Observed 2026-11-21 · source benchlm:models |
| 27 | [A.X K2](https://modelchoice.dev/models/a-x-k2) | SK Telecom | 75.9 | Observed 2026-11-21 · source benchlm:models |
| 28 | [Qwen3.6 Plus](https://modelchoice.dev/models/qwen3-6-plus) | Alibaba | 75.8 | Observed 2026-11-21 · source benchlm:models |
| 28 | [Spark-X2.6-4B](https://modelchoice.dev/models/spark-x2-6-4b) | SparkLLM | 75.8 | Observed 2026-11-21 · source benchlm:models |
| 30 | [LFM2.5-24B-A2B](https://modelchoice.dev/models/lfm2-5-24b-a2b) | LiquidAI | 74.69 | Observed 2026-11-21 · source benchlm:models |
| 31 | [Ling 3.0 Flash](https://modelchoice.dev/models/ling-3-0-flash) | InclusionAI | 74.5 | Observed 2026-11-21 · source benchlm:models |
| 32 | [Granite 4.2 3B](https://modelchoice.dev/models/granite-4-2-3b) | IBM | 74.33 | Observed 2026-11-21 · source benchlm:models |
| 33 | [Nemotron 3 Nano Omni 30B A3B](https://modelchoice.dev/models/nemotron-3-nano-omni-30b-a3b) | NVIDIA | 74.2 | Observed 2026-11-21 · source benchlm:models |
| 34 | [Ternary Bonsai 2 27B](https://modelchoice.dev/models/ternary-bonsai-2-27b) | Prism ML | 74 | Observed 2026-11-21 · source benchlm:models |
| 35 | [Ling 3.0 Flash FP8](https://modelchoice.dev/models/ling-3-0-flash-fp8) | InclusionAI | 73.4 | Observed 2026-11-21 · source benchlm:models |
| 36 | [LFM3-1.7B](https://modelchoice.dev/models/lfm3-1-7b) | LiquidAI | 73.03 | Observed 2026-11-21 · source benchlm:models |
| 37 | [Nemotron 3.5 Lightning 30B A3B NVFP4](https://modelchoice.dev/models/nemotron-3-5-lightning-30b-a3b-nvfp4) | NVIDIA | 72.88 | Observed 2026-11-21 · source benchlm:models |
| 38 | [K-EXAONE 2.0](https://modelchoice.dev/models/k-exaone-2-0) | LG AI Research | 72.6 | Observed 2026-11-21 · source benchlm:models |
| 39 | [Nemotron 3.5 Super VL 120B A12B NVFP4](https://modelchoice.dev/models/nemotron-3-5-super-vl-120b-a12b-nvfp4) | NVIDIA | 71.5 | Observed 2026-11-21 · source benchlm:models |
| 40 | [Nemotron 3.5 Super VL 120B A12B](https://modelchoice.dev/models/nemotron-3-5-super-vl-120b-a12b) | NVIDIA | 70.86 | Observed 2026-11-21 · source benchlm:models |
| 41 | [Agents-A1-4B](https://modelchoice.dev/models/agents-a1-4b) | InternScience | 69.1 | Observed 2026-11-21 · source benchlm:models |
| 42 | [MiniCPM5-2B](https://modelchoice.dev/models/minicpm5-2b) | OpenBMB | 66.3 | Observed 2026-11-21 · source benchlm:models |
| 43 | [Hy3 Preview](https://modelchoice.dev/models/hy3-preview) | Tencent | 63.1 | Observed 2026-11-21 · source benchlm:models |
| 44 | [LFM2.5-2.6B](https://modelchoice.dev/models/lfm2-5-2-6b) | LiquidAI | 59.17 | Observed 2026-11-21 · source benchlm:models |
| 45 | [Claude Opus 4.5](https://modelchoice.dev/models/claude-opus-4-5) | Anthropic | 58 | Observed 2026-11-21 · source benchlm:models |
| 46 | [Ling 2.6 Flash](https://modelchoice.dev/models/ling-2-6-flash) | InclusionAI | 57 | Observed 2026-11-21 · source benchlm:models |
| 47 | [LFM2.5-8B-A1B](https://modelchoice.dev/models/lfm2-5-8b-a1b) | LiquidAI | 56.47 | Observed 2026-11-21 · source benchlm:models |
| 48 | [OrcaCyber-Nano-1.0](https://modelchoice.dev/models/orcacyber-nano-1-0) | OrcaRouter | 56.33 | Observed 2026-11-21 · source benchlm:models |
| 49 | [Solar Pro 3](https://modelchoice.dev/models/solar-pro-3) | Upstage | 55.78 | Observed 2026-11-21 · source benchlm:models |
| 50 | [ZAYA1-8B](https://modelchoice.dev/models/zaya1-8b) | Zyphra | 52.56 | Observed 2026-11-21 · source benchlm:models |
| 51 | [Trinity-Large-Thinking](https://modelchoice.dev/models/trinity-large-thinking) | Arcee AI | 52.3 | Observed 2026-11-21 · source benchlm:models |
| 52 | [MiniCPM5-1B](https://modelchoice.dev/models/minicpm5-1b) | OpenBMB | 46.67 | Observed 2026-11-21 · source benchlm:models |
| 53 | [LFM2.5-230M](https://modelchoice.dev/models/lfm2-5-230m) | LiquidAI | 38.4 | Observed 2026-11-21 · source benchlm:models |
| 54 | [Kanana-2 1.3B Instruct](https://modelchoice.dev/models/kanana-2-1-3b-instruct) | Kakao | 34.69 | Observed 2026-11-21 · source benchlm:models |
| 55 | [Kanana-2 3B Instruct](https://modelchoice.dev/models/kanana-2-3b-instruct) | Kakao | 33.33 | Observed 2026-11-21 · source benchlm:models |
| 56 | [LFM2.5-VL-3B](https://modelchoice.dev/models/lfm2-5-vl-3b) | LiquidAI | 25.8 | Observed 2026-11-21 · source benchlm:models |
| 57 | [LLaDA2.2-mini](https://modelchoice.dev/models/llada2-2-mini) | InclusionAI | 24.93 | 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.
