Forge is available: build a website with AI from your RodiumAi account.

Try Forge
Back to all models
DeepSeek

DeepSeek

Open-weight Chinese frontier models — outstanding cost/quality with thinking and non-thinking modes.

Available models

3

Starting rate

175.8 RODI / 1M

Max context

1M

DeepSeek
DeepSeek V4 Prodeepseek/deepseek-v4-pro
chatopen-sourcereasoninglong-contexttoolsfrontier

DeepSeek V4 Pro via Microsoft Foundry on Azure. Azure azure_ai/DeepSeek-V4-Pro. Most capable DeepSeek V4 MoE model with 1M context; supports non-thinking and thinking modes (thinking default), tool calls, and JSON output.

Price

  • In527.2RODI/M~0.660 USD/M
  • Out1581.6RODI/M~1.98 USD/M
  • Cached17.6RODI/M~0.0220 USD/M
Context1MSpeedDeepInput:textOutput:textModel details
DeepSeek
DeepSeek V4 Flashdeepseek/deepseek-v4-flash
chatopen-sourcereasoninglong-contexttools

DeepSeek V4 Flash via Microsoft Foundry on Azure. Azure azure_ai/DeepSeek-V4-Flash. General-purpose MoE flagship with 1M context; supports non-thinking and thinking modes (thinking default), tool calls, and JSON output.

Price

  • In175.8RODI/M~0.220 USD/M
  • Out527.2RODI/M~0.660 USD/M
  • Cached5.6RODI/M~0.0070 USD/M
Context1MSpeedFastInput:textOutput:textModel details
DeepSeek
DeepSeek V3.2deepseek/deepseek-v3.2
chatcodingopen-sourceserverless

DeepSeek V3.2 is a mixture-of-experts model for reasoning, coding, and instruction following. Text in, text out, with streaming, tool use, and structured outputs.

Price

  • In467.2RODI/M~0.620 USD/M
  • Out1394.1RODI/M~1.85 USD/M
Context164KSpeedBalancedInput:textOutput:textModel details