DeepSeek V3.2
DeepSeek V3.2 is a large language model made by DeepSeek, added to the catalog on 1 Dec 2025. It has a context window of 163,840 tokens and costs $0.26 per 1M input tokens and $0.38 per 1M output tokens. It accepts text and produces text. Typical uses: enhance reasoning in complex tasks, support long-context interactions and facilitate agentic tool-use scenarios.
Price checked · source: OpenRouter
What DeepSeek V3.2 is
DeepSeek V3.2, released by deepseek on December 1, 2025, is a powerful text-to-text model designed for high computational efficiency and strong reasoning abilities. With an impressive context length of 131,072 tokens, this model excels in long-context scenarios. It employs a unique DeepSeek Sparse Attention (DSA) mechanism that optimizes training and inference costs while maintaining quality. The model also features a scalable reinforcement learning framework that enhances its reasoning capabilities, placing it in the GPT-5 performance class. Users can easily control reasoning behavior, making it suitable for various interactive environments. With its ability to synthesize agentic tasks, DeepSeek V3.2 improves compliance and generalization, making it a versatile choice for developers and researchers alike.
Key features
- DeepSeek Sparse Attention for efficiency
- 131,072 token context length
- Scalable reinforcement learning framework
- Control over reasoning behavior
- High performance in interactive environments
Typical uses
- Enhance reasoning in complex tasks
- Support long-context interactions
- Facilitate agentic tool-use scenarios
- Improve compliance in AI applications
DeepSeek V3.2 pricing
US dollars, list price through OpenRouter.
| Item | Standard |
|---|---|
| Input, per 1M tokens | $0.26 |
| Output, per 1M tokens | $0.38 |
| Cached input read, per 1M tokens | $0.13 |
No price change since it was added on 1 Dec 2025.
| What a task costs | Cost |
|---|---|
| Answer a short question500 in / 300 out | $0.0002 |
| Summarise a 50-page PDF25,000 in / 800 out | $0.0068 |
| Review a 2,000-line source file30,000 in / 2,000 out | $0.0086 |
| Read a 300-page contract set150,000 in / 1,500 out | $0.040 |
DeepSeek V3.2 specifications
| Model id | deepseek/deepseek-v3.2 |
|---|---|
| Version | deepseek-v3.2-20251201 |
| Context window | 163,840 tokensAbout 122,880 English words |
| Max output | 147,456 tokens |
| Knowledge cutoff | Not published |
| Added to catalog | 1 Dec 2025 |
| Input types | text |
|---|---|
| Output types | text |
| Tool calling | Yes |
| Structured outputs | Yes |
| Reasoning effort control | No |
| Content moderation | Not applied by provider |
Sample outputs
The same prompts are given to every model, and the unedited output is shown with the date it was produced. Upend does not publish leaderboard scores it did not run.
Write a short essay
RUN20 Nov 2026Write 150 words on how AI might positively upend work, leisure and creativity
AI is poised to positively upend our lives by transforming labor, leisure, and creativity. In the workplace, it will automate routine tasks, freeing humans to focus on complex problem-solving, strategic thinking, and empathetic roles. This shift could foster more meaningful careers and a better work-life balance. For leisure, AI will offer hyper-personalized experiences, from curating entertainment to acting as a tireless tutor for new hobbies, making personal growth more accessible. Most profoundly, AI is emerging as a collaborative partner in creativity. It can act as a muse for artists, a co-writer for authors, and a compositional tool for musicians, democratizing creative expression and pushing the boundaries of human imagination. By handling the mundane, AI empowers us to become more human—more innovative, more connected, and more fulfilled.