---
id: "deepseek/deepseek-v4-pro"
also: ["deepseek/deepseek-v4-pro-20260423", "deepseek-ai/DeepSeek-V4-Pro", "hf/deepseek-ai/DeepSeek-V4-Pro", "requesty/deepseek/deepseek-v4-pro", "vercel/deepseek/deepseek-v4-pro"]
card_usd_per_m: "0.66 in, 1.98 out, at deepseek"
cheapest_list_usd_per_m: "0.66 in, 1.98 out, at deepseek on openrouter"
cheapest_effective_usd_per_m: "0.1878 in, 1.9098 out, at streamlake/fp8 on openrouter"
context: 1048576
max_output: 384000
input: ["text"]
output: ["text"]
released: "2026-04-24"
reasoning: ["xhigh", "high", "none"]
api: ["/api/v1/chat/completions", "/api/v1/completions", "/api/v1/messages", "/api/v1/responses"]
params: ["frequency_penalty", "include_reasoning", "logit_bias", "logprobs", "max_completion_tokens", "max_tokens", "min_p", "presence_penalty", "reasoning", "reasoning_effort", "repetition_penalty", "response_format", "seed", "stop", "structured_outputs", "temperature", "tool_choice", "tools", "top_k", "top_logprobs", "top_p"]
endpoints: 47
gateways: "openrouter (23), vercel (11), requesty (7), hf (6)"
no_training: 36
trains: 1
retains_prompts: 11
intelligence: 12.6000
---
# DeepSeek: DeepSeek V4 Pro 0423

DeepSeek V4 Pro is DeepSeek's flagship Mixture of Experts model (1.6T total, 49B active parameters) with a 1M token context window, built for advanced reasoning, coding, and long-horizon agentic workflows with strong tool use. Served by Lyceum in the EU.