Qwen: Qwen Plus 0728 (thinking)
Model Type
Proprietary Model
API access only
Recommended Use Cases
Try Qwen Plus 0728 (thinking)
Qwen-Plus-0728 (thinking) is the reasoning-enabled variant of Alibaba's balanced proprietary model, with 1M token context and extended chain-of-thought for complex tasks requiring deliberate reasoning.
Qwen-Plus supports both thinking and non-thinking modes. You can switch between them using the enable_thinking parameter.
- Alibaba Cloud
Overview
Qwen-Plus-0728 (thinking) is the same model as Qwen-Plus-0728 with thinking mode enabled by default. It produces visible reasoning traces before generating final answers, improving accuracy on math, coding, logic, and complex multi-step problems.
Key Features
- Extended reasoning: Visible
<think>blocks for complex problems - 1M context: Million-token context window
- Strong reasoning: Significantly outperforms QwQ on benchmarks
- Tool integration: Accurate tool invocation during reasoning
- Human alignment: Enhanced for complex instruction following
Technical Specifications
| Specification | Value |
|---|---|
| Architecture | Proprietary (based on Qwen3) |
| Context Length | 1M tokens |
| Thinking Mode | Enabled by default |
| Release Date | September 2025 |
| License | Proprietary (API only) |
Thinking vs Non-Thinking
| Aspect | Thinking (this model) | Non-thinking |
|---|---|---|
| Response style | Chain-of-thought + answer | Direct answer |
| Latency | Higher | Lower |
| Token usage | Higher | Lower |
| Accuracy on hard tasks | Higher | Lower |
| Cost | Higher | Lower |
When to Use Qwen-Plus-0728 (thinking)
Choose thinking mode when you need:
- Complex mathematical reasoning
- Multi-step logical problems
- Code debugging and analysis
- Tasks benefiting from step-by-step reasoning
- Maximum accuracy over speed
Choose non-thinking when you need:
- Fast, direct responses
- Simple queries
- Cost optimization
- Production with tight latency requirements
Pricing
- Input: $0.40/M tokens
- Output: $4.00/M tokens
Higher than non-thinking due to reasoning token consumption.
Availability
- API: Alibaba Cloud Model Studio, OpenRouter
- Web: chat.qwen.ai (with thinking toggle)
- Open Weights: Not available (proprietary)
Role in Series
Qwen-Plus variants:
- Qwen-Plus-0728: Balanced, hybrid mode (default non-thinking)
- Qwen-Plus-0728 (thinking): Reasoning-focused (this model)