Model Library
DeepSeekReleased December 1, 2025DeepSeek V3.2 Speciale
DeepSeek · Released December 1, 2025
Intelligence score
22
At a glance
- Intelligence
- 22
- Speed
- —
- Latency
- —
- Input price
- Not published
- Output price
- Not published
- Context window
- 130Ktokens
- Parameters
- 685billions
Summary
Comparison summary
DeepSeek V3.2 Speciale scores 22 (estimated) on the Artificial Analysis Intelligence Index, placing it below average among other open weight models of similar size (median: 25).
Specifications
Technical specifications
| Model name | DeepSeek V3.2 Speciale |
|---|---|
| Provider | DeepSeek |
| Release date | December 1, 2025 |
| Intelligence score | 22 |
| Speed | — |
| Latency (TTFT) | — |
| Input price | Not published |
| Output price | Not published |
| Context window | 130,000 tokens |
| Parameters | 685B |
| Reasoning | Yes |
| Input modalities | text |
| Output modalities | — |
| Open weights | Yes |
| License | MIT |
| API availability | — |
| Source | Model Intelligence legacy profile: deepseek-v3-2-speciale |
FAQ · 16
Frequently asked questions
DeepSeek V3.2 Speciale was released on December 1, 2025.
DeepSeek V3.2 Speciale was created by DeepSeek.
DeepSeek V3.2 Speciale scores 22 (estimated) on the Artificial Analysis Intelligence Index, placing it below average among other open weight models of similar size (median: 25).
Yes, DeepSeek V3.2 Speciale is a reasoning model. It uses extended thinking or chain-of-thought reasoning to work through complex problems before providing an answer.
DeepSeek V3.2 Speciale supports text input.
DeepSeek V3.2 Speciale supports text output.
No, DeepSeek V3.2 Speciale does not support image input. It can only process text.
No, DeepSeek V3.2 Speciale is not multimodal. It only supports text input.
DeepSeek V3.2 Speciale has a context window of 130k tokens. This determines how much text and conversation history the model can process in a single request.
Yes, DeepSeek V3.2 Speciale is open weights. The model weights are publicly available and can be downloaded for self-hosting.
DeepSeek V3.2 Speciale has 685 billion parameters (37 billion active).
DeepSeek V3.2 Speciale is a Mixture of Experts (MoE) model with 685 billion total parameters, but only 37 billion active parameters are used during inference.
DeepSeek V3.2 Speciale is released under the MIT license. This license allows commercial use.
DeepSeek V3.2 Speciale achieves a score of 22 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding.
DeepSeek V3.2 Speciale is an open weights model that can be self-hosted.
DeepSeek V3.2 Speciale is an open weights model that can be downloaded and self-hosted.