DeepSeek releases V4.1 Flash 1017 update with MIT-licensed weights and unchanged API pricing
Article timeline
The articles in this story. Times are ET.
vLLM@vllm_projectOfficialvLLM reports DeepSeek-V4.1-Flash-1017 serving throughput and cost on SemiAnalysis AgentXAIvLLM congratulates DeepSeek on DeepSeek-V4.1-Flash-1017, which the post says has better scores and shorter outputs for long agent loops. vLLM says it delivers 1.3B tokens per dollar TCO on SemiAnalysis AgentX, at 830K tok/s per GPU (cached + input + output) at 100 tok/s per user. The model is live on DeepSeek's website, app, and API, and the post links a vLLM blog on the work.
Thomas Wolf@Thom_WolfXDeepSeek releases V4.1-Flash-1017 with higher Terminal-Bench 4.0 and DeepSWE scoresAIDeepSeek released DeepSeek-V4.1-Flash-1017, which it describes as a Pareto improvement with stronger benchmarks, faster task completion, and shorter outputs. Thomas Wolf highlights Terminal-Bench 4.0 rising from 31.2 to 42.3 and DeepSWE at 77.0, versus 74.0 for Claude Opus 5 and 73.0 for GPT-5.6 Sol. He notes the weights are MIT-licensed and questions whether the numbers will hold outside DeepSeek's own eval setup.
wh@nrehiew_XDeepSeek-V4.1-Flash-1017 claimed Pareto gains in performance and costAIDeepSeek says DeepSeek-V4.1-Flash-1017 outperforms its previous version on benchmarks, completes tasks faster, and produces shorter outputs. The model is now live on DeepSeek's website, app, and API, and is optimized for DeepSeek Harness v0.2.1-rc.1 or later.
Orange AI@oran_geXDeepSeek releases V4.1-Flash-1017 with stronger benchmarks and shorter outputsAIDeepSeek says its DeepSeek-V4.1-Flash-1017 model outperforms the previous V4.1-Flash on benchmarks, completes tasks faster, and produces shorter outputs. The model is available on DeepSeek's website, app, and API, and is optimized for DeepSeek Harness and Tencent CodeBuddy. The author says it cuts overthinking and saves about half the tokens on the same tasks.
Vercel Developers@vercel_devOfficialVercel adds DeepSeek V4.1 Flash 1017 to AI GatewayAIVercel says DeepSeek V4.1 Flash 1017 is now available on AI Gateway, with gains in reasoning, coding, agentic tasks, and images. The model is also said to finish tasks faster and use tokens more efficiently.
Guizang@op7418XDeepSeek releases V4.1 Flash 1017 with a DeepSeek SWE V1.1 benchmark gain and a search APIAIDeepSeek has released V4.1 Flash 1017, which Guizang says significantly improves overall capability while shortening task run time and output length at lower cost. The release emphasizes building complete projects from scratch with simple prompts and no manual intervention, and DeepSeek also opened its in-house search API.
🚨 AI News | TestingCatalog@testingcatalogXDeepSeek releases open-weight V4.1-Flash-1017, scoring 77% on DeepSWE 1.1AIDeepSeek released DeepSeek-V4.1-Flash-1017, an open-weight model that scores 77% on the DeepSWE 1.1 benchmark, according to X: Testing Catalog. The post says the model outperforms GLM-5.3, Kimi K3, GPT-5.6 Sol and Opus 5 on that benchmark and is available through DeepSeek Chat, its API and HuggingFace. DeepSeek's own post says the model is faster, produces shorter outputs and improves cost-performance over the previous version.
X.PIN@thexpinXDeepSeek opens its app search engine to developers at RMB 5 per 1,000 callsAIDeepSeek is opening the search engine behind its own app to developers through new Search and Fetch API endpoints. The endpoints cost RMB 5 (about $0.70) per 1,000 calls and filter out SEO spam and low-quality AI content. The post also says V4.1 Flash received a 1017 update with faster, cheaper runs and open weights.
OpenRouter@OpenRouterOfficialDeepSeek V4.1 Flash 1017 is live on OpenRouterAIOpenRouter has added DeepSeek V4.1 Flash 1017, with DeepSeek as the first provider and more providers planned. DeepSeek says the update is more token efficient and shows major gains on agentic benchmarks.
DeepSeek@deepseek_aiOfficialDeepSeek releases V4.1-Flash-1017 with better benchmarks and lower costAIDeepSeek says DeepSeek-V4.1-Flash-1017 delivers stronger benchmark performance, faster task completion, and shorter outputs than the previous version. The company calls the change a Pareto improvement in performance and cost. The model is live on DeepSeek's website, app, and API.
- OpenRouter · New modelsBlogDeepSeek releases V4.1 Flash 1017, an October 17 snapshot of V4.1 Flash
AIDeepSeek says V4.1 Flash 1017 is its October 17, 2026 snapshot of DeepSeek V4.1 Flash, served under the deepseek-flash API model name. V4.1 Flash is a sparse mixture-of-experts model.
Ivan Fioravanti@ivanfioravantiXDeepSeek releases open-weight DeepSeek-V4.1-Flash-1017 on Hugging FaceAIDeepSeek has released DeepSeek-V4.1-Flash-1017 with open weights on Hugging Face. The post gives no benchmark scores, parameter counts, or context length.
- DeepSeek API NewsOfficialPickDeepSeek upgrades V4.1 Flash to DeepSeek-V4.1-Flash-1017
AIDeepSeek has upgraded its V4.1 Flash model to DeepSeek-V4.1-Flash-1017, with web and app models updated too. The API pricing and calling method are unchanged, and developers access it by using deepseek-flash. The company reports gains across benchmarks, including 77.0 on DeepSWE v1.1 and 42.6 on HLE without tools, alongside shorter task runtime and output length.
- DeepSeek · new models on Hugging FaceOfficialPickDeepSeek releases DeepSeek-V4.1-Flash-1017, a minor update to V4.1-Flash
AIDeepSeek releases DeepSeek-V4.1-Flash-1017 on Hugging Face as a minor update to DeepSeek-V4.1-Flash. The source says it improves reasoning, coding, agentic and visual benchmarks while reducing task execution time and output length. The model's weights are released under the MIT License with a 1M-token context window.