DeepSeek-V3.2

Name: DeepSeek-V3.2
Author: DeepSeek

DeepSeekOpen WeightsPending Human Review

DeepSeek-V3.2 combines DeepSeek Sparse Attention with scalable reinforcement learning and large-scale synthetic agent training. It supports both ordinary responses and explicit reasoning, including tool use within thinking mode. The release emphasizes mathematical reasoning, software engineering, and general agent workflows while reducing long-context inference cost. Public MIT weights permit self-hosting. DeepSeek originally exposed it through the deepseek-chat and deepseek-reasoner aliases, which subsequently migrated to V4; the model card and weights identify the specific release. Speciale, the separate high-compute reasoning specialization, has its own page.

2025-12-01

671B total, 37B activated

Mixture of Experts (MoE) with DeepSeek Sparse Attention

MIT

Compare with other models

Specifications

Parameters: 671B total, 37B activated
Architecture: Mixture of Experts (MoE) with DeepSeek Sparse Attention
License: MIT
Context Window: 128,000 tokens
Type: text
Modalities: text

Benchmark Scores

Advanced Specifications

Model Family: DeepSeek
API Access: Not Available
Chat Interface: Not Available

Capabilities & Limitations

Capabilities: reasoningcode generationmathematicsreasoning with toolsagentic coding
Known Limitations: Open-weight deployment requires substantial memory and model-specific inference support.Reasoning and tool-based benchmark results depend on the evaluation harness and sampling configuration.Historical API aliases have moved to newer models.
Notable Use Cases: coding assistantsmathematical problem solvingdocument analysistool-using agents
Function Calling Support: Yes
Tool Use Support: Yes

Resources

Related Models

DeepSeek-V4.1-Flash

DeepSeek

DeepSeek-V4.1-Flash is a natively multimodal MoE model trained from scratch on 45 trillion tokens. It uses a causal encoder-decoder layout, Compressed Sparse Attention 2, Engram conditional memory, and DSpark speculative decoding to reduce inference and KV-cache cost. Its backbone activates different parameter counts during prefill and decoding, while a separate 196B conditional-memory table is sparsely accessed. The model accepts images and text and generates text, supports million-token contexts and adjustable reasoning effort, and powers the official deepseek-flash API. MIT weights and model-specific inference code support self-hosting.

Typemultimodal

Parameters552B backbone + 196B Engram memory; 8B active prefill / 16B active decode

2026-09-10

Open Weights

Details Compare

DeepSeek-V4-Flash-Vision-Exp

DeepSeek

DeepSeek-V4-Flash-Vision-Exp is the experimental vision-language specialization of the production V4 Flash model. It adds image understanding for agent tasks while retaining the text capabilities of V4 Flash, enabling workflows that inspect interfaces, diagrams, and other visual context. It was publicly accessible through a dedicated DeepSeek API model name before downloadable MIT weights were released. Following the September 10 V4.1 Flash launch, the experimental endpoint was retired and its alias redirected; the original checkpoint remains available for self-hosted research.

DeepSeek-V4-Flash

DeepSeek

DeepSeek-V4-Flash is the smaller, efficiency-focused tier of the V4 family, first publicly released as a preview on April 24, 2026. Its hybrid Compressed Sparse Attention and Heavily Compressed Attention architecture supports million-token contexts alongside manifold-constrained hyper-connections. It supports non-thinking and thinking modes for code, knowledge, and agent tasks. The 2026-07-31 production update (DeepSeek-V4-Flash-0731) adds substantially stronger agent behavior and DSpark speculative decoding, and is included as a named checkpoint variant. The original Flash API has retired and its alias now routes to V4.1 Flash; original and updated V4 weights remain downloadable.

Typetext

Parameters284B total, 13B activated

2026-04-24

Open Weights

Details Compare