DeepSeek-V4.1-Flash-1017
Multimodal 1M-context MoE checkpoint for coding, reasoning, and tool-using agents
Available via compatible API
Available with Zero Data RetentionDeepSeek-V4.1-Flash-1017
Multimodal 1M-context MoE checkpoint for coding, reasoning, and tool-using agents
Available via compatible API
Available with Zero Data RetentionDeepSeek-V4.1-Flash-1017 Overview
DeepSeek-V4.1-Flash-1017 is the October 2026 checkpoint of DeepSeek-V4.1-Flash, a multimodal mixture-of-experts model for coding, reasoning, visual understanding, and tool-using agents. It reads text and images, writes text, and combines a 1M-token context window and up to 1M output tokens with continuously adjustable reasoning effort and tool calls. It suits large-codebase work, document analysis, research, and long-running agentic tasks.