DeepSeek
DeepSeek V3.2
DeepSeek chat model for instruction following, coding, and analysis
Model facts
- Context window
- 164K tokens
- Maximum output
- 66K tokens
- Cheapest paid input
- $0.18 per 1M tokens
- Cheapest paid output
- $0.35 per 1M tokens
- Open weights
- yes
- Providers
- 37
Capabilities and modalities
reasoning, tool calling, structured output, attachments, temperature control, open weights, text, image, pdf.
Local hardware estimate
At 4K context: Q4_K_M 420.5 GiB, Q8_0 739.6 GiB, F16 1338.1 GiB working memory.
- Parameters
- 685.4 billion
- Architecture
- deepseekv32
- Model license
- mit
Planning estimate adapted from the MIT-licensed whichllm estimator using metadata from Hugging Face; not a vendor minimum requirement.
Artificial Analysis benchmarks
Compare this model on the full LLM benchmark leaderboard.
| Configuration | AAI | Coding | Math | Output tokens/s |
|---|---|---|---|---|
| Reasoning | 32.8 | 44.2 | 92.0 | not measured |
| Non-reasoning | 25.1 | not measured | 59.0 | not measured |
Observed price history
- 2026-08-24: $0.18 input / $0.35 output per 1M tokens
API providers
- Alibaba Token Plan — model id deepseek-v3.2; free input / free output per 1M tokens; provider documentation
- Alibaba Token Plan (China) — model id deepseek-v3.2; free input / free output per 1M tokens; provider documentation
- SCNet Token Plan — model id DeepSeek-V3.2; free input / free output per 1M tokens; provider documentation
- CrofAI — model id deepseek-v3.2; $0.18 input / $0.35 output per 1M tokens; provider documentation
- TokenGo — model id deepseek/deepseek-v3.2; $0.217 input / $0.326 output per 1M tokens; provider documentation
- Deep Infra — model id deepseek-ai/DeepSeek-V3.2; $0.26 input / $0.38 output per 1M tokens; provider documentation
- DevPass (LLM Gateway) — model id deepseek-v3.2; $0.26 input / $0.38 output per 1M tokens; provider documentation
- LLM Gateway (DeepInfra) — model id deepinfra/deepseek-v3.2; $0.26 input / $0.38 output per 1M tokens; provider documentation
- Meganova — model id deepseek-ai/DeepSeek-V3.2; $0.26 input / $0.38 output per 1M tokens; provider documentation
- OpenRouter — model id deepseek/deepseek-v3.2; $0.269 input / $0.4 output per 1M tokens; provider documentation
- Kilo Gateway — model id deepseek/deepseek-v3.2; $0.269 input / $0.4 output per 1M tokens; provider documentation
- LLM Gateway (NovitaAI) — model id novita/deepseek-v3.2; $0.269 input / $0.4 output per 1M tokens; provider documentation
- NovitaAI — model id deepseek/deepseek-v3.2; $0.269 input / $0.4 output per 1M tokens; provider documentation
- Abacus — model id deepseek-ai/DeepSeek-V3.2; $0.27 input / $0.4 output per 1M tokens; provider documentation
- Poe — model id novita/deepseek-v3.2; $0.27 input / $0.4 output per 1M tokens; provider documentation
- Helicone — model id deepseek-v3.2; $0.27 input / $0.41 output per 1M tokens; provider documentation
- Hugging Face — model id deepseek-ai/DeepSeek-V3.2; $0.28 input / $0.4 output per 1M tokens; provider documentation
- LLM Gateway (ByteDance) — model id bytedance/deepseek-v3.2; $0.28 input / $0.42 output per 1M tokens; provider documentation
- NanoGPT — model id deepseek/deepseek-v3.2; $0.28 input / $0.42 output per 1M tokens; provider documentation
- Vercel AI Gateway — model id deepseek/deepseek-v3.2; $0.28 input / $0.42 output per 1M tokens; provider documentation
- Vivgrid — model id deepseek-v3.2; $0.28 input / $0.42 output per 1M tokens; provider documentation
- ZenMux — model id deepseek/deepseek-v3.2; $0.28 input / $0.43 output per 1M tokens; provider documentation
- 302.AI — model id deepseek-v3.2; $0.29 input / $0.43 output per 1M tokens; provider documentation
- Ofox — model id deepseek/deepseek-v3.2; $0.29 input / $0.43 output per 1M tokens; provider documentation
- Merge Gateway — model id deepseek/deepseek-v3.2; $0.28 input / $0.45 output per 1M tokens; provider documentation
- Cortecs — model id deepseek-v3.2; $0.296 input / $0.495 output per 1M tokens; provider documentation
- TensorX — model id deepseek/deepseek-v3.2; $0.3 input / $0.5 output per 1M tokens; provider documentation
- Venice AI — model id deepseek-v3.2; $0.33 input / $0.48 output per 1M tokens; provider documentation
- Friendli — model id deepseek-ai/DeepSeek-V3.2; $0.5 input / $1.5 output per 1M tokens; provider documentation
- Vultr — model id nvidia/DeepSeek-V3.2-NVFP4; $0.55 input / $1.65 output per 1M tokens; provider documentation
- LLM Gateway (Vertex AI (OpenAI-compatible)) — model id vertex-openai/deepseek-v3.2; $0.56 input / $1.68 output per 1M tokens; provider documentation
- Vertex — model id deepseek-ai/deepseek-v3.2-maas; $0.56 input / $1.68 output per 1M tokens; provider documentation
- Azure — model id deepseek-v3.2; $0.58 input / $1.68 output per 1M tokens; provider documentation
- Azure Cognitive Services — model id deepseek-v3.2; $0.58 input / $1.68 output per 1M tokens; provider documentation
- EmpirioLabs AI — model id deepseek-v3-2; $0.57 input / $1.71 output per 1M tokens; provider documentation
- Amazon Bedrock — model id deepseek.v3.2; $0.62 input / $1.85 output per 1M tokens; provider documentation
- Qiniu — model id deepseek/deepseek-v3.2-251201; not published input / not published output per 1M tokens; provider documentation