Skip to main content
leaderboard
.cn
Toggle Sidebar
Search models or organizations…
⌘K
Explore
Frontier Models
Historical Trends
Local Models
Compare Models
Ranking Studio
Library
Models
Pricing
Organizations
Benchmarks
Data sources
Settings
Toggle Sidebar
leaderboard
.cn
Loading page content
Feedback
Loading page content
Back to models
Qwen3-4B-Instruct-2507
Open
Share
Add to comparison
Tool use
Overview
Deployment
Evaluations
Evaluation overview
Overall rank
#129
59.9 points · 180 models
Open-weight rank
#60
66.1 points · 109 models
Data coverage
25 items
2 sources · updated through 2025-07-01
Domain performance
Coding
#115 / 160
64.1
Research
#122 / 174
70.5
Agents
#105 / 125
56.2
Evaluation highlights
All evaluations
MultiPL-E
Coding · Qwen3-4B-Instruct-2507 Technical Report
76.8%
3
/ 31
ZebraLogic
Reasoning · Qwen3-4B-Instruct-2507 Technical Report
80.2%
12
/ 39
BFCL v3 Full
Agents · Qwen3-4B-Instruct-2507 Technical Report
61.9%
27
/ 49
IFEval
Instruction following · Qwen3-4B-Instruct-2507 Technical Report
83.4%
24
/ 30
Creative Writing v3
Writing · Qwen3-4B-Instruct-2507 Technical Report
83.5%
8
/ 42
MultiIF
Multilingual · Qwen3-4B-Instruct-2507 Technical Report
69%
17
/ 45
Basic information
Context (input / output)
262K / 16K
Parameters
4B
Release date
2025-07
License
Apache 2.0
Input modalities
Text
Output modalities
Text
Modes
Instruct / non-thinking only
Version
2507
Availability
SGLang · FP8 · 125K
Deployment options
FP8
Related links
Official site
HuggingFace
Version relationships
Qwen3-4B family
Qwen3-4B
2025-04 · Same series
Qwen3-4B-Base
2025-04 · Same series
Qwen3-4B-Instruct-2507
Current
2025-07 · Current model
Qwen3-4B-Thinking-2507
2025-07 · Same series