Identity
- Status
- active
- Developer ID
deepseek-ai/DeepSeek-V3.2- Released
- Dec 1, 2025
- Version
- DeepSeek-V3.2
- License
- mit
DeepSeek-V3.2 is an open-weight language model published by DeepSeek.
| Provider | Input / 1M | Output / 1M | Cache read / 1M | Cache write / 1M | Observed | Evidence |
|---|---|---|---|---|---|---|
DeepinfraAvailable | $0.26 | $0.38 | $0.13 | — | Oct 19, 2026 | Provider-reported |
OpenrouterAvailable | $0.269 | $0.40 | $0.1345 | — | Aug 28, 2026 | Provider-reported |
| LMArena Text Arenatext-2026-10-15-a64ca3b12b6e | 1,424.46 | — | 48,112 | deepseek-v3.2; 95% CI [1420.99052972, 1427.92299771]; votes 48112; rank 105unknown | Original sourcewinner eligible | Third-party benchmark |
At a Glance
deepseek-ai/DeepSeek-V3.2This page represents one developer model product. Dated API snapshots, serving endpoints, pricing, regions, quantizations, and service tiers attach as versioned aliases or provider details and never create duplicate public model records.
Primary Evidence
Comparable Peers
| A | Pair | B | Context |
|---|---|---|---|
DeepSeek V3.2DeepSeek | vs | Phi-4 ReasoningMicrosoft | cross-developer peerstext |
DeepSeek V3.2DeepSeek | vs | DeepSeek V4 ProDeepSeek | family variantstext |
DeepSeek V3.2DeepSeek | vs | DeepSeek V4 Pro BaseDeepSeek | family variantstext |
DeepSeek V3.2DeepSeek | vs | DeepSeek R1DeepSeek | family variantstext |
DeepSeek V3.2DeepSeek | vs | DeepSeek V3DeepSeek | family variantstext |
DeepSeek V3.2DeepSeek | vs | Llama 3.1 405BMeta | cross-developer peerstext |
DeepSeek V3.2DeepSeek | vs | cross-developer peerstext | |
DeepSeek V3.2DeepSeek | vs | Llama 3.1 70BMeta | cross-developer peerstext |
DeepSeek V3.2DeepSeek | vs | Kimi K2 ThinkingMoonshot AI | cross-developer peerstext |
DeepSeek V3.2DeepSeek | vs | GLM 5Z.ai | cross-developer peerstext |
DeepSeek V3.2DeepSeek | vs | GLM 5.1Z.ai | cross-developer peerstext |
DeepSeek V3.2DeepSeek | vs | GLM 5.2Z.ai | cross-developer peerstext |
Questions
DeepSeek-V3.2 is an open-weight language model published by DeepSeek. It is developed by DeepSeek and its current sourced lifecycle status is active.
DeepSeek V3.2's current record lists Text as input and Text as output.
The current record lists a 164K-token context window and does not report a maximum output length.
DeepSeek V3.2 is marked as API-available. The directly sourced provider records currently include Deepinfra, Hugging Face, and Openrouter.
The current record lists 685.4B total parameters and an unknown active parameter count. Its architecture is DeepseekV32ForCausalLM.
DeepSeek V3.2's recorded capabilities are chat, generation, and reasoning. Its supported tasks are text.
The lowest directly sourced prices currently attached to DeepSeek V3.2 are $0.26 per million input tokens through Deepinfra and $0.38 per million output tokens through Deepinfra. Prices are provider-specific and should be checked against each cited observation date.
DeepSeek V3.2 is an open-weight model. Self-hosting is supported, and the recorded license is mit.
No predecessor is recorded. No successor is recorded. Recorded aliases are deepseek-ai/DeepSeek-V3.2.
The current compatible peer set includes Phi-4 Reasoning, DeepSeek V4 Pro, DeepSeek V4 Pro Base, DeepSeek R1, and DeepSeek V3. Compatibility requires shared sourced modalities and tasks; it does not imply that one model is better. Browse model comparisons →