A DeepSeek model for text generation and reasoning tasks that need more capability.
Use deepseek/deepseek-v4-pro through the sjolt LLM API.
deepseek/deepseek-v4-proToken rates
deepseek/deepseek-v4-proSimilar Models
Model overview
Take on complex reasoning, production agentic systems, and coding work that benefits from a higher-capability model and deliberate analysis.
Give production agents room to plan, call tools, inspect results, and revise complex implementations across longer chains of work.
Bring large codebases, technical records, requirements, and prior decisions into a 1M-token context window for connected analysis.
Integrate with OpenAI Chat Completions, OpenAI Responses, or Anthropic Messages while keeping one sjolt endpoint.
Model characteristics
Complex reasoning, production agentic systems, and demanding coding tasks.
Up to 1,000,000 input tokens.
Up to 384,000 tokens.
Disable thinking when it is not needed, or select low, high, or max effort.
FAQ
Choose it for complex reasoning, production agentic workflows, or coding tasks where additional model capability is more important than the lighter Flash profile.
Yes. It is positioned for multi-step agents that plan, use tools, evaluate results, and refine substantial coding or operational work.
The API supports up to 1M context tokens and up to 384K output tokens.
Thinking can be disabled. When enabled, set effort to low, high, or max to match the task.