DeepSeek V3.2 Exp
DeepSeek-V3.2-Exp is an experimental large language model developed by DeepSeek that serves as an intermediate release between version 3.1 and future architectures. It features a large context window and utilizes a fine-grained sparse attention mechanism known as DeepSeek Sparse Attention to process text inputs.
DeepSeek V3.2 Exp specs and pricing
- Provider: DeepSeek
- Released: September 29, 2025
- Context window: 164K tokens
- Max output: 147K tokens
- Input: text
- Output: text
- Price: $0.27 input / $0.41 output per 1M tokens
- Reasoning: yes
- Tool calling: yes
What DeepSeek V3.2 Exp is good at
- Complex reasoning tasks requiring deep analysis
- Processing long-form text documents
- Applications requiring integrated tool calling capabilities
To-dos to try with DeepSeek V3.2 Exp
- Analyze a lengthy research paper and summarize the core methodology.
- Use tool calling to retrieve specific data points from a provided text dataset.
- Draft a detailed technical report based on a set of provided notes.
FAQ
- What is the primary innovation in this model? This model introduces DeepSeek Sparse Attention, which is a fine-grained sparse attention mechanism designed to improve processing efficiency.
- Does this model support non-text inputs? No, the model is currently limited to text input and text output.
- How much does DeepSeek V3.2 Exp cost? DeepSeek V3.2 Exp costs $0.27 per million input tokens and $0.41 per million output tokens through OpenRouter. In todo.is a typical agent step costs about 0.81 credits ($1 = 100 credits).
- What is the context window of DeepSeek V3.2 Exp? DeepSeek V3.2 Exp can read up to 164K tokens at once and write up to 147K tokens in one reply.
- Can I use DeepSeek V3.2 Exp without coding? Yes. In todo.is you pick DeepSeek V3.2 Exp as your agent's model and write to-dos in plain words, on the web, the Android app, WhatsApp or Telegram. No API key needed: credits are included, or you can bring your own key.
JavaScript is required to use the todo.is app.