DeepSeek updates its open-weight V4.1 Flash model to finish tasks faster with shorter answers

DeepSeek released V4.1-Flash-1017, saying it scores better on benchmarks, finishes tasks faster and writes shorter outputs than the previous version, improving performance and cost at once. It's live in DeepSeek's app and API, and the weights are free to download on Hugging Face. DeepSeek also opened its web Search and Fetch APIs to everyone at $0.75 per 1,000 calls.

What it means for manufacturers and fabricators

Agents that run all day, chasing quotes or keying in POs, pay for every word they write, so shorter answers and faster finishes cut the cost per job. Open weights also mean a company can host it on its own servers if it doesn't want shop data leaving the building.

Source: DeepSeek

All AI news · Practical AI guides on our blog

DeepSeek updates its open-weight V4.1 Flash model to finish tasks faster with shorter answers

What it means for manufacturers and fabricators

Also worth knowing

Hyundai and Kia show Polaris, a factory AI agent that answers spoken questions about defects with causes and fixes

China's LEVEL AI turns one machining drawing into a quote, process card, schedule and inspection sheet, and open-sources its model

Edmonton's SketchDeck.ai launches SketchDeck 3.0, adding connection material and drawing-revision automation to its AI steel takeoff tool

Meta releases Muse Spark Open, an open-weight model for coding and agent work that can run on two Mac Studios