instruction following benchmark
IFBench Leaderboard
IFBench is an Instruction Following Benchmark evaluating a model's ability to follow complex instructions.
Updated Sep 5, 2026
Models39
Model coverage39
MetricScore
EvidenceB
IFBench Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score82.80% | Percentile100.00% | Participants39 | EvidenceC | Evaluated |
| Rank02 | ModelTM | Score82.20% | Percentile97.37% | Participants39 | EvidenceC | Evaluated |
| Rank03 | ModelNV | Score81.70% | Percentile94.74% | Participants39 | EvidenceC | Evaluated |
| Rank04 | ModelAC | Score81.30% | Percentile92.11% | Participants39 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score81.30% | Percentile89.47% | Participants39 | EvidenceC | Evaluated |
| Rank06 | ModelNR | Score81.21% | Percentile86.84% | Participants39 | EvidenceC | Evaluated |
| Rank07 | ModelAM Nova 2 ProAmazon | Score80.20% | Percentile84.21% | Participants39 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score79.50% | Percentile81.58% | Participants39 | EvidenceC | Evaluated |
| Rank09 | ModelIB | Score79.33% | Percentile78.95% | Participants39 | EvidenceC | Evaluated |
| Rank10 | ModelAC | Score79.10% | Percentile76.32% | Participants39 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score79.10% | Percentile73.68% | Participants39 | EvidenceC | Evaluated |
| Rank12 | ModelIB | Score77.17% | Percentile71.05% | Participants39 | EvidenceC | Evaluated |
| Rank13 | ModelME | Score77.00% | Percentile68.42% | Participants39 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score76.50% | Percentile65.79% | Participants39 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score76.50% | Percentile63.16% | Participants39 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score76.10% | Percentile60.53% | Participants39 | EvidenceC | Evaluated |
| Rank17 | ModelMI | Score75.00% | Percentile57.89% | Participants39 | EvidenceC | Evaluated |
| Rank18 | ModelIB | Score74.33% | Percentile55.26% | Participants39 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score74.20% | Percentile52.63% | Participants39 | EvidenceC | Evaluated |
| Rank20 | ModelCO | Score74.00% | Percentile50.00% | Participants39 | EvidenceC | Evaluated |
| Rank21 | ModelNV | Score72.56% | Percentile47.37% | Participants39 | EvidenceC | Evaluated |
| Rank22 | ModelNV | Score71.88% | Percentile44.74% | Participants39 | EvidenceC | Evaluated |
| Rank23 | ModelIN | Score71.00% | Percentile42.11% | Participants39 | EvidenceC | Evaluated |
| Rank24 | ModelAMNova 2 LiteAmazon | Score70.80% | Percentile39.47% | Participants39 | EvidenceC | Evaluated |
| Rank25 | ModelAC | Score70.20% | Percentile36.84% | Participants39 | EvidenceC | Evaluated |
| Rank26 | ModelMI | Score70.00% | Percentile34.21% | Participants39 | EvidenceC | Evaluated |
| Rank27 | ModelOP | Score69.50% | Percentile31.58% | Participants39 | EvidenceC | Evaluated |
| Rank28 | ModelMI | Score69.00% | Percentile28.95% | Participants39 | EvidenceC | Evaluated |
| Rank29 | ModelMA | Score69.00% | Percentile26.32% | Participants39 | EvidenceC | Evaluated |
| Rank30 | ModelAMNova 2 OmniAmazon | Score68.70% | Percentile23.68% | Participants39 | EvidenceC | Evaluated |
IFBench Highlights
The leading models and scores on this benchmark.
IFBench Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for IFBench
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is IFBench?
What IFBench measures and how its scores work.
IFBench is an Instruction Following Benchmark.
It measures a model's ability to follow complex instructions using the Score metric, reported as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- IFBench
- Modality
- text
- Primary category
- instruction following
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about IFBench.
Which model scores highest on IFBench?
Qwen3.8 Max is currently ranked first with 82.80%.
What are the top three models on IFBench?
The current leaders are Qwen3.8 Max (82.80%), Inkling-Small (82.20%), and Nemotron 3 Ultra (550B A55B) (81.70%).
Which IFBench model has the lowest official input price?
Qwen3.8 Flash has the lowest matched official input price at $0.15 input / $0.47 output per 1M tokens.
Which models are fastest among IFBench results?
The fastest matched records are Mercury 2 (346.33 tok/s via Inception), Mistral Small 4 (163.71 tok/s via Mistral AI), and MiniMax M2.1 (100.00 tok/s via MiniMax).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.