Loading page content
Qwen3-30B-A3BOpen
Tool use · Reasoning
Evaluation overview
Overall rank
#109
64.3 points · 180 models
Open-weight rank
#44
71.3 points · 109 models
Data coverage
114 items
4 sources · updated through 2026-07-16
Domain performance
Coding#113 / 160
64.6
Research#100 / 174
75.8
Agents#95 / 125
59
Long context#79 / 82
52.5
Evaluation highlights
All evaluationsCoding · Qwen3-4B-Instruct-2507 Technical Report · Initial April 2025 model · non-reasoning mode
74.6%
5/ 31
Instruction following · Qwen3-4B-Instruct-2507 Technical Report · Initial April 2025 model · non-reasoning mode
83.7%
23/ 30
Basic information
- Context (input / output)
- 131K / 33K
- Parameters
- 30.5B / 3.3B active
- Release date
- 2025-04
- License
- Apache 2.0
- Input modalities
- Text
- Output modalities
- Text
- Modes
- Supports thinking and non-thinking modes
- Version
- Initial release, April 2025
- Active parameters
- 3.3B
Availability
SGLang · FP8 · 125KDeployment options
FP8
Related links
Version relationships
Qwen3-30B-A3B family
Qwen3-30B-A3BCurrent
2025-04 · Current model
Qwen3-30B-A3B-Base
2025-04 · Same series
Qwen3-30B-A3B-Instruct-2507
2025-07 · Same series
Qwen3-30B-A3B-Thinking-2507
2025-07 · Same series