DeepSeek releases V4.1-Flash-1017 with open weights and opens its Search and Fetch APIs
Announced · Published · By Contractor AI Shop
DeepSeek released V4.1-Flash-1017, a minor update to V4.1-Flash, on its website, app and API, with weights on Hugging Face. DeepSeek claims better coding and agent benchmark scores with shorter outputs, and opened Search and Fetch APIs to all users at $0.75 per 1,000 calls.
Key facts
- DeepSeek describes V4.1-Flash-1017 as a minor update that improves reasoning, coding, agentic and visual benchmarks while reducing task time and output length. DeepSeek
- The Decoder reports the update beats larger rivals on agentic tasks while cutting per-task costs; benchmark figures are DeepSeek's own. The Decoder
- MarkTechPost reports a DeepSWE score of 77.0 and that DeepSeek opened its Search API to all users. MarkTechPost
Our take: what it means for contractors
A cheap, strong hosted option for drafting and agent work, but at hundreds of billions of parameters it is not a model for shop hardware. Think about where client documents go before sending them to a China-hosted API; U.S. hosting providers are an alternative.
What we don't know yet
- Independent benchmark results
- Memory needs for usable quantized builds
- Exact license terms on the weights
Sources
- DeepSeek · DeepSeek-V4.1-Flash-1017 · 2026-10-17 · Primary
- The Decoder · Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs · 2026-10-17 · Secondary
- MarkTechPost · DeepSeek Just Released V4.1-Flash-1017, Scores 77.0 on DeepSWE, Opens Search API · 2026-10-17 · Secondary