Loading page content
Qwen3.8-Flash-NextOpen
Tool use · Reasoning
Evaluation overview
Overall rank
#16
90.2 points · 197 models
Open-weight rank
#5
92.4 points · 118 models
Data coverage
27 items
1 source · updated through 2026-08-26
Domain performance
Coding#18 / 177
86.2
Research#26 / 191
90.7
Agents#10 / 143
88.6
Vision#6 / 78
89.8
Evaluation highlights
All evaluationsBasic information
- Context (input / output)
- 1M / 66K
- Parameters
- 125B / 6B active
- Release date
- 2026-08
- License
- Qwen Community
- Input modalities
- Text · Image · Video
- Output modalities
- Text
- Positioning
- Preview of the Qwen4 model architecture
- Architecture
- MoE · Gated DeltaNet + QSA Hybrid
- Modes
- Supports thinking and non-thinking modes
- Total parameters
- 125B
- Exact release date
- 2026-08-26
- API pricing
- ¥1 input / ¥3 output per million tokens (API coming soon)
- Weight status
- Open weights released (Qwen Community License)
- API model ID
- qwen3.8-flash
- N-gram embedding parameters
- 51B
- Default reasoning effort
- xhigh (supports low / medium / xhigh)