Deepseek releases V4.1-Flash-1017, beats larger rivals in agentic tasks and cuts per-task costs by over 60%

Deepseek has released an updated Flash model, V4.1-Flash-1017, which outperforms its predecessor and larger rivals like Claude Opus 5 and GPT 5.6-Sol in coding and agentic tasks. The model achieves higher benchmark scores while reducing average per-task costs to under $0.20, a significant drop from the predecessor's $0.50. Deepseek is also opening its internal search engine to outside developers via public Search & Fetch APIs, priced at $0.75 per 1,000 calls.