DeepSeek releases V4.1-Flash-1017, an open MIT-licensed update with a 1M context

Monday, 19 October 2026By Sridhar Mukkandi1 min read

DeepSeek published DeepSeek-V4.1-Flash-1017 on Hugging Face on 17 October under the MIT licence, describing it as a minor update to V4.1-Flash with improvements in reasoning, coding, agentic tasks and visual benchmarks, and shorter outputs that lower cost. The model has 763 billion parameters, takes images and text, and supports a 1-million-token context.

On DeepSeek's own results it scores 42.3 on Terminal-Bench 4.0, up from 31.2 for V4.1-Flash, and 77.0 on DeepSWE v1.1, up from 74.2. DeepSeek recommends a temperature of 1.0 and a maximum output of at least 256,000 tokens.

So what

An MIT licence means you can run and modify it commercially with few strings attached. The scores are DeepSeek's own, and shorter outputs only save money if quality holds on your tasks, so test both before you switch.

SourceDeepSeek on Hugging Facehuggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017 Go deeper · EssayYour LLM's 0.95 isn't a probability

Drafted with AI from the original source, then checked and edited by Sridhar Mukkandi. Spotted a mistake? Write to hello@tarkika.com and we'll correct it here.

More from The Signal

Research Microsoft’s ThinkingBox turns database checks into rewards for training agentsChecking the state an agent leaves behind is a more honest test than reading its transcript, and the same check doubles as a training reward. Start with the three actions your agent takes most often. Microsoft and Hugging Face Agents Bayer Crop Science runs 22 read-only agents over its ML decision engine on AWSRead-only agents with narrowly scoped tools, launched to a small group first, is a pattern worth copying. So is testing a new model against your own tool calls before you switch: newer isn't always compatible. AWS Machine Learning Blog Tools AWS lets you write your own model-training loop on SageMaker without managing GPUsCustom reinforcement learning used to mean renting and babysitting GPUs. If you can write a check for a good outcome, you can now train on it without running the infrastructure yourself. AWS Machine Learning BlogAll stories