Instruction Following

IFEval Leaderboard

Instruction Following Evaluation

Why This Matters

Reliability in following instructions - crucial for automation and production systems

Good Scores

70%+ is reliable, 80%+ is very dependable, 85%+ is production-ready

Use Cases

  • •Automated workflows
  • •Code generation
  • •Document formatting
  • •API integrations

Peak Score

89.98

Average

45.64

Models Tested

53,438

Median Score

45.48

Efficiency Leaders

Best performance per billion parameters - The smart choices

#1

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#2

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#3

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#4

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#5

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#6

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#7

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#8

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#9

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

#10

ChatWaifu_v1.4

100.0M params • Score: 56.91

Efficiency

569.06

Performance by Model Size

How different size classes perform on this benchmark

small

< 1B

Avg Score: 23.44

ChatWaifu_v1.456.91
ChatWaifu_v1.456.91
ChatWaifu_v1.456.91

medium

1B - 10B

Avg Score: 48.46

virtuoso-small-v2-tensopolis-v184.19
virtuoso-small-v2-tensopolis-v184.19
virtuoso-small-v2-tensopolis-v184.19

large

10B - 50B

Avg Score: 51.80

ZYH-LLM-Qwen2.5-14B-V385.78
ZYH-LLM-Qwen2.5-14B-V385.78
ZYH-LLM-Qwen2.5-14B-V385.78

xlarge

> 50B

Avg Score: 64.33

Llama-3.3-70B-Instruct89.98
Llama-3.3-70B-Instruct89.98
Llama-3.3-70B-Instruct89.98

🏆 Open Source Champions

Top permissively licensed models

ZYH-LLM-Qwen2.5-14B-V3

license:apache-2.0

85.78

ZYH-LLM-Qwen2.5-14B-V3

license:apache-2.0

85.78

ZYH-LLM-Qwen2.5-14B-V3

license:apache-2.0

85.78

ZYH-LLM-Qwen2.5-14B-V3

license:apache-2.0

85.78

ZYH-LLM-Qwen2.5-14B-V3

license:apache-2.0

85.78

📈 Most Downloaded Models

Popularity meets performance

gpt2

11.8M downloads • Rank #48183

17.93

gpt2

11.8M downloads • Rank #48183

17.93

gpt2

11.8M downloads • Rank #48183

17.93

gpt2

11.8M downloads • Rank #48183

17.93

gpt2

11.8M downloads • Rank #48183

17.93

📄 License Analysis

Performance by license type

llama89.98
20447 modelsAvg: 47.08
Unknown86.62
14487 modelsAvg: 45.65
license:apache-2.085.78
14019 modelsAvg: 44.58
license:mit84.16
2592 modelsAvg: 41.51
base_model:ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.273.48
46 modelsAvg: 58.40
license:gpl-3.072.04
168 modelsAvg: 25.42
license:cc-by-nc-sa-4.069.25
46 modelsAvg: 61.70
license:cc-by-nc-4.067.45
1019 modelsAvg: 48.59

🔧 Framework Analysis

Performance by framework

OTHER89.98
53299 modelsAvg: 45.67
PYTORCH56.98
116 modelsAvg: 39.80
HUGGINGFACE22.66
23 modelsAvg: 22.66

About IFEval

IFEval tests a model's ability to follow specific instructions precisely, including formatting requirements, length constraints, and structural specifications. Critical for real-world application reliability.

Last updated: 11/25/2025

Test These Models Yourself

Run benchmarks on your own data with these platforms

Together.ai

Instant API access to this model

Fastest API

Production-ready inference API. Start free, scale to millions.

Try Free API

Modal

Run this model on serverless GPU

Most Popular

Deploy in seconds with $30 free credits. Pay only for what you use.

Get $30 Free Credits

RunPod

Rent GPU starting at $0.34/hour

Best Value

Deploy on cloud GPU or serverless. 70% cheaper than AWS.

Start from $0.34/hr

Disclosure: We may earn a commission from these partners. This helps keep LLMYourWay free.

Complete Leaderboard

Top 50 models ranked by IFEval performance

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#4

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#5

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#6

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#7

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#8

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#9

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#10

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#11

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#12

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#13

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#14

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#15

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#16

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#17

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#18

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#19

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#20

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#21

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#22

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#23

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#24

Llama-3.3-70B-Instruct

meta-llama• llama

89.98

Score

#25

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#26

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#27

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#28

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#29

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#30

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#31

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#32

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#33

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#34

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#35

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#36

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#37

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#38

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#39

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#40

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#41

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#42

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#43

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#44

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#45

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#46

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#47

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#48

Llama-3.1-70B-Instruct

meta-llama• llama

86.69

Score

#49

calme-2.1-qwen2.5-72b

MaziyarPanahi

86.62

Score

#50

calme-2.1-qwen2.5-72b

MaziyarPanahi

86.62

Score