DeepSeek V4.1 Flash 1017 listed on OpenRouter

The model has 763.2 billion parameters and a context window of 1,048,576 tokens. It is open weights and was released on 2026-10-17. Its Hugging Face ID is deepseek-ai/DeepSeek-V4.1-Flash-1017. No free variant is available.

deepseek-ai announced deepseek-ai/DeepSeek-V4.1-Flash-1017. According to deepseek-ai, DeepSeek-V4.1-Flash-1017 is a minor update to DeepSeek-V4.1-Flash. deepseek-ai says DeepSeek-V4.1-Flash-1017 delivers substantial improvements across reasoning, coding, agentic tasks, and visual benchmarks. deepseek-ai says The model reduces task execution time and output length. deepseek-ai says The model enables users to complete tasks faster and at lower cost. deepseek-ai says The model advances the performance–cost Pareto frontier across multiple agent harnesses. According to deepseek-ai, HLE w/o tools score for DS-V4.1-Flash-1017 is 42.6 (44.0†). deepseek-ai says Terminal-Bench 4.0 score for DS-V4.1-Flash-1017 is 42.3. deepseek-ai says DeepSWE v1.1 score for DS-V4.1-Flash-1017 is 77.0. deepseek-ai says ProgramBench score for DS-V4.1-Flash-1017 is 29.1. deepseek-ai says NL2Repo-Bench score for DS-V4.1-Flash-1017 is 68.2. deepseek-ai says Automation-Bench v1.0.6 score for DS-V4.1-Flash-1017 is 60.5. deepseek-ai says Agents’ Last Exam score for DS-V4.1-Flash-1017 is 35.4.