The extended-reasoning (thinking) variant of the July 2025 Qwen3 30B-A3B release, producing explicit chain-of-thought for better performance on reasoning and multi-step tasks while remaining cheap to run.
Cost rate
0.4x
Context
82K
Released
Aug 28, 2025
Input
Text
Output
Text
Support
ReasoningTool callingStructured outputs
Performance
Median latency and throughput measured across recent requests.
Throughput109 tok/s
Latency2.2s
Latest news about Qwen3 30B A3B Thinking 2507
Recent articles and community posts, linked to their original sources.
Mapping every neuron in Qwen3-8BHacker News372- Qwen3-Coder writes less secure code when told it's for a US personaHacker News243118
- Show HN: Qwen3-4B with the AI writing tics trained outHacker News158113

I Tested Qwen 4 Max on Mac: It’s Actually FasterBruno Vega4,612
Your Qwen 3.8 27B Just Got 3 Free UpgradesRepoChad15,911
I Gave Qwen 4 Max and Fable 5.2 the Same Impossible PromptBetter Stack44,867
Qwen 4 Max is here! OpenAI and Anthropic are in troubleSimon Høiberg20,232