DeepSeek 4.1-1017 nearly matches Opus 5
DeepSeek released 4.1-1017, an upgraded model for coding, agents, 3D and vision that nearly matches Opus 5 while remaining cheap and open-weight.
DeepSeek has released 4.1-1017, a model that, according to the developers, nearly matches Opus 5. This is a significant step forward compared with DeepSeek 4.1 Flash: the model handles 3D, programming, agentic tasks and visual information better.
Key highlights:
- Terminal-Bench: 42.8
- SOTA on DeepSWE
- Outperforms all Chinese models while remaining very cheap
- Completes tasks faster and generates shorter answers
- API prices unchanged, context window of 1 million tokens
The model is already available in the DeepSeek app, the web version and via API. The weights are published on Hugging Face. For local deployment, two NVIDIA DGX Spark systems can be used.
Why it matters
AnalysisIt matters because DeepSeek shows that frontier AI models can combine strong coding and agent performance with low API costs and open weights.
Discuss in community
Share your questions and insights with developers