Alibaba QwenLanguage modelSupersededOpen weights
Qwen3 235B
Apache-licensed hybrid reasoning family that became the default open base model.
Overall Intelligence
#51
of 56
26.3 pts behind #1
0
races led
14
weeks at #1
62.4
score
- Released
- Apr 29, 2025
- 1.5 yr ago
- Context
- 128K
- 32K max output
- Price
- $0.2 / $0.6
- per 1M in/out · $0.30 blended
- Speed
- 70 tok/s
- 1s to first token
- Modalities
- text
- outputs text
- Weights
- Open
- 235B MoE (22B active)
The 30-second version
Apache-licensed hybrid reasoning family that became the default open base model.
Key takeaways
- 1Apache-licensed hybrid reasoning family that became the default open base model.
Where it ranks
Current snapshot · illustrative scores · click any race to watch it
Qwen3Best on each axis
Overall
Coding
51Coding+27.5 pts51Frontend & UI+170 Elo51Backend & APIs+24.0 pts49Distributed Systems+26.9 pts51Full-Stack Apps+27.1 pts50Debugging & Repair+27.7 pts51Agentic Coding+29.1 pts49Code Review & Security+24.4 pts52Data Science & SQL+26.0 pts50Mobile Apps+23.7 pts51Terminal & DevOps+31.6 pts48Competitive Programming+768
Hard Reasoning
Math
Science
Writing & Language
Multimodal
Tool Use & Agents
Overall Intelligence over time
#1 frontierQwen3V4.1 FlashGLM-5.3
Update history
The distilled version first; open any release for the deeper breakdown.
The Qwen Max family