Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs
DeepSeek shipped open-weight V4.1-Flash-1017, claiming better agentic coding at under $0.20 per task.
DeepSeek released DeepSeek-V4.1-Flash-1017 with open weights and API access, reporting higher scores, faster runs, and shorter outputs than V4.1-Flash. On DeepSWE it scores 77.0% at just under $0.20 per task, or about 72% near $0.10, versus 74.2% at roughly $0.50 for the predecessor. DeepSeek says it beats larger models including Claude Opus 5, GPT-5.6-Sol, Kimi K3, and GLM-5.3 on several coding and agent tests, though independent benchmarks are not yet available. It also opened its in-house search engine via Search and Fetch APIs at $0.75 per 1,000 calls.
73