DeepSeek-V4-Pro

DeepSeek-V4-Pro is DeepSeek's flagship reasoning model, exposed with the official catalog id `DeepSeek-V4-Pro`. CallMissed accepts `"model": "DeepSeek-V4-Pro"` on `/v1/chat/completions`. Legacy lowercase aliases may still resolve, but new integrations should use the catalog id exactly as published.\n\nCapabilities: a full one-million (1,048,576) token context window, function calling, reasoning ("thinking") mode, and vision (image input). It is built for long-horizon agentic workflows, large codebases, and multi-step problem-solving that exceed the context limits of most other hosted models.\n\nPricing on CallMissed is $1.32 per million input tokens and $3.96 per million output tokens, with cached input billed at $0.044 per million tokens. For teams that want frontier-class reasoning plus a 1M context and native tool calling, V4 Pro is the top DeepSeek tier.\n\nUse DeepSeek-V4-Pro for codebase analysis, long-document review, tool-using agents, and multimodal tasks that combine text and images. Reasoning content can increase time-to-first-token; stream responses to keep UIs responsive, and set `max_tokens` generously so a long thinking phase does not truncate the answer.\n\nIntegration is OpenAI-compatible: send chat messages, define `tools` for function calling, and include `image_url` parts for vision. If your client library surfaces a separate reasoning field, handle it the same way you would for other thinking models. When it profiles as overkill for easy traffic, route simpler prompts to `DeepSeek-V4-Flash`.