DeepSeek V4.1-Flash-1017: MIT Weights, Built-in Search and Fetch APIs, Higher Agent Benchmarks
What is it?
A minor update to September's V4.1-Flash. DeepSeek says it raises reasoning, coding, agent and vision scores while cutting task time and output length, which lowers cost. It has a 1M-token context and the weights are MIT-licensed. In the API, use the name deepseek-flash; the old deepseek-v4-flash name now routes to this model. The same day DeepSeek opened Search and Fetch APIs built on its own search engine, plus a developer preview of an agent harness.
Why does it matter?
By DeepSeek's own model-card numbers, Terminal-Bench 4.0 went from 31.2 to 42.3Vendor claim and HLE without tools from 36.8 to 42.6Vendor claim. Teams building agents can plug in the same vendor's search and fetch without a separate service. Old model names now route to the new model automatically, so check whether your results shift.
Maturity
GABoth the API and the MIT weights are out. There are no independent measurements yet, and no inference provider hosts it on Hugging Face so far.
How to track it
- Independent tests: watch for V4.1-Flash-1017 scores on Artificial Analysis and whether they match DeepSeek's numbers.
- Pricing: check the Models & Pricing page in the DeepSeek API docs for Search and Fetch rates and any model price change.
- Next model: follow DeepSeek news for a
deepseek-v4-prorefresh or a V4.1-Pro.