DeepSeek V3.2
#17 in Open-Source-Spraakmodelledeepseek · v3.2 · siet 2025-12-01 · 30× · tolest 30. Juni 2026
34
Momentum
DeepSeek-V3.2 is an open-weight language model based on a Mixture-of-Experts (MoE) architecture with 685 billion total parameters, of which only 37 billion are activated per token. The model introduces DeepSeek Sparse Attention (DSA), a new attention mechanism that significantly reduces computational overhead for long-context scenarios. Released on December 1, 2025 under the MIT license, it supports integrated "thinking" reasoning directly within tool-use mode and substantially narrows the performance gap between open-source and closed frontier models.
Momentum-Verloop
04.04.03.07.
Features
| Context Window (Tokens) | 163,840 tokens (max context length; per official tech report: 128K native, extended to ~163K) |
| License | MIT License (model weights and code) |
| Multimodality | Text only (no native image/audio/video input); dedicated multimodal model is DeepSeek-VL2 (separate product) |
| Platform | DeepSeek Web, App, API (deepseek-chat); model weights on Hugging Face & GitHub; supported inference frameworks: SGLang, vLLM (Day-0 support); also available on Microsoft Azure Foundry |
| Price | Free to use via DeepSeek Web & App; API: from $0.028/1M input tokens (cache hit) to $0.28/1M input tokens (cache miss), $0.42/1M output tokens |
| Price per 1M Tokens | $0.28 / 1M input tokens (cache miss); $0.028 / 1M input tokens (cache hit, 90% discount); $0.42 / 1M output tokens – via DeepSeek Official API |
| Release Date | December 1, 2025 |