AITOPAI热报

DeepSeek V4.1 Flash

DeepSeek·China·Released 2026-10-17
Open weightsGeneral LLMReasoningTool useVision

DeepSeek's large language model supporting text and image input, excels at reasoning and agentic coding with a 1M token context window.

1Mcontext
$0.15 / $0.6per 1M tokens (in / out)
763Bparameters
291stories in our feed

Specs

hover a value to see its source
Context window1M
Max output384K
Parameters763B
Knowledge cutoff2025-05
Licencemit
Commercial useCommercial OK
InputText · Image
OutputText
HF downloads (30d)1.9M
Familydeepseek-flash

API price

USD per 1M tokens
Input$0.15
Output$0.6
Cached input$0.003
Change history
2026-10-20 Context 1M → 1Mmodels.dev

News timeline

from the AITOP feed, one story per eventAll 120 →
2026-11-05
Science's Last Exam benchmarks AI research agents on reproducible workrohanpaul_ai
2026-11-03
J-PACT Detects Collusion Between LLM Agents via J-SpacearXiv: DeepSeek
2026-10-31
OpenCode 启动 Cheepseek 第四阶段：10 美元订阅享 100 美元 DeepSeek V4.1 Flash 额度IT之家
2026-10-29
豆包工作向企业客户开放第三方模型：接入 Kimi 3.1 Pro、GLM 5.3 等IT之家
2026-10-28
DeepSeek V4.1 Flash tops FrontierCode 2.0 among open-weight modelsIT之家
2026-10-24
vLLM sets best DeepSeek-V4.1-Flash result on SemiAnalysis AgentXvLLM
2026-10-20
开源动态第25期：DeepSeek V4.1 Flash 1017、美国本土开源模型与中美监管新规Nathan Lambert: Interconnects
DeepSeek 开源推理框架 DeepInfer，Apache 2.0 协议发布marktechpost
DeepSeek 宣布将内部推理框架 DeepInfer 全面开源IT之家
2026-10-19
DeepSeek V4.1 Flash 1017 上架 Databricks，推理与编程能力提升Databricks
MiniMax M3.1 Flash 半价追平 DeepSeek V4.1 Flash 1017Decoder
2026-10-17
Ollama 开始向 deepseek-v4.1-flash 用户推送更新后的模型ollama
2026-10-15
AMD MI355X joins AA-AgentPerf benchmark against NVIDIA BlackwellArtificial Analysis
2026-10-14
Gloss：基于Together AI的开源应用Together AI
2026-10-12
Agent Arena 数据：开源模型 12 周追平 Claude Fable 5lmarena.ai
2026-10-10
Tsinghua, Oxford, Stanford Paper Finds LLMs Reason Even When Told Not Torohanpaul_ai
Lenovo's TianxiCode Agent and DeepSeek-V4.1-Flashpandaily
2026-10-09
Claude Pro Opus 5.5与DeepSeek V4.1 Flash每 token成本相同Teortaxes
2026-10-08
vLLM Highlights DeepSeek-V4.1-Flash SWA OptimizationvLLM
2026-10-07
Microsoft 演示 DeepSeek V4 Flash 本地跑在 Windows PC 上rohanpaul_ai
vLLM 三周优化 DeepSeek-V4.1-Flash：低并发提速 1.9 倍vLLM
2026-10-01
Qwen3.8-Max 修复 PVE 服务器故障Geek
URAI Robot-Agent InterfacearXiv: DeepSeek
2026-09-29
Neuro-Symbolic Reasoning for Autonomous DrivingarXiv: DeepSeek
2026-09-28
博主用 DeepSeek V4.1 Flash 做 RSS 翻译，每天花费不到 1 元向阳乔木
2026-09-27
VS Code 装 Cline 插件可免费用 Gemini 3.8 Flash、DeepSeek V4.1 Flash 等模型向阳乔木
2026-09-26
Jev 占 OpenRouter 分类请求量 27%，超过 DeepSeek V4 FlashOpenRouter
OpenCode Makes US$60 DeepSeek V4.1 Flash Credit PermanentIT之家
2026-09-24
Gemini 3.8 Flash 在 Cline 中免费可用Cline
单文件 Minecraft 测试：DeepSeek V4.1 Flash 胜过 Space Bunny Alpha@atomic_chat_hq
Zenith 系统让 DeepSeek v4.1 Flash 在长程任务上超过 GPT 5.6 SolEmad Mostaque
2026-09-23
Bolt Forge 一周数据：GLM 5.3 Flash 与 DeepSeek V4.1 Flash 占八成请求bolt.new
Goldman Sachs: US Outspends China on AI Compute $806bn vs $110bnAI Will
GPT-6-Luna 与 DeepSeek-V4.1-Flash 价格对比：各有优势区间Geek
2026-09-20
OpenAI 用 1 万个代理展示纳维-斯托克斯方程失效，DeepSeek V4.1 Flash 发布新架构DeepLearning.AI
DeepSeek V4.1-Flash model ships 1M context with extreme KV compressionpandaily
2026-09-19
Notion 支持使用 DeepSeek V4.1 Flash 模型Notion
智谱GLM 5.3 FlashX性能提升但价格更高，被DeepSeek v4.1 Flash超越Browser Use
2026-09-18
Bolt 上开源模型首周，开发者选最快 DeepSeek V4.1 Flashbolt.new
2026-09-15
DeepSWE测试显示DeepSeek V4.1-Flash输入token占比远高于输出Fireworks AI