{"full_name":"deepseek-ai/DeepInfer","owner":"deepseek-ai","name":"DeepInfer","shard":"1","url":"https://github.com/deepseek-ai/DeepInfer","description":"DeepInfer: High-performance, large-scale inference framework for DeepSeek models","homepage":null,"topics":[],"tag_pages":[],"language":"Python","license":"Apache-2.0","avatar":"https://avatars.githubusercontent.com/u/148330874?v=4","created_at":"2026-10-16T07:47:23Z","pushed_at":"2026-11-03T07:31:59Z","metrics":{"stars":26094,"forks":3207,"watchers":64,"issues":0,"prs":251,"releases":0,"issues_open":0,"issues_closed":0,"prs_open":225,"prs_closed":24,"prs_merged":2,"discussions":9},"growth":{"day":{"stars":42,"forks":9,"base_day":"2026-11-15"},"week":{"stars":382,"forks":55,"base_day":"2026-11-09"}},"history":{"dates":["2026-10-24","2026-10-25","2026-10-26","2026-10-27","2026-10-28","2026-10-29","2026-10-30","2026-10-31","2026-11-01","2026-11-02","2026-11-03","2026-11-04","2026-11-09","2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[11937,17600,19763,20960,21760,22977,23365,23867,24220,24476,25016,25182,25712,25778,25854,25895,25931,25991,26052,26094],"forks":[1060,1799,2104,2311,2473,2671,2731,2831,2898,2950,3036,3063,3152,3162,3169,3177,3181,3187,3198,3207],"watchers":[26,38,43,47,51,55,56,55,57,60,64,64,65,65,64,64,64,64,64,64],"issues_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,0,0,0,0,0,0],"prs_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,221,221,221,224,224,225]},"boards":{"monthly":37,"monthly_rate":32,"new":11},"same_owner":[{"full_name":"deepseek-ai/deepseek-harness","name":"deepseek-harness","shard":"1","boards":{"daily":43,"weekly":21,"monthly":7,"daily_rate":192,"monthly_rate":200,"daily_forks":26,"daily_watchers":17,"total":17}},{"full_name":"deepseek-ai/DeepSeek-V3","name":"DeepSeek-V3","shard":"1","boards":{"total":134}},{"full_name":"deepseek-ai/DeepSeek-R1","name":"DeepSeek-R1","shard":"1","boards":{"total":175}},{"full_name":"deepseek-ai/3FS","name":"3FS","shard":"1","boards":{}},{"full_name":"deepseek-ai/DeepGEMM","name":"DeepGEMM","shard":"1","boards":{}},{"full_name":"deepseek-ai/DeepSeek-OCR","name":"DeepSeek-OCR","shard":"1","boards":{}},{"full_name":"deepseek-ai/DeepSpec","name":"DeepSpec","shard":"1","boards":{}},{"full_name":"deepseek-ai/FlashMLA","name":"FlashMLA","shard":"1","boards":{}},{"full_name":"deepseek-ai/Janus","name":"Janus","shard":"1","boards":{}},{"full_name":"deepseek-ai/awesome-deepseek-agent","name":"awesome-deepseek-agent","shard":"1","boards":{}},{"full_name":"deepseek-ai/awesome-deepseek-integration","name":"awesome-deepseek-integration","shard":"1","boards":{}},{"full_name":"deepseek-ai/chat-demo","name":"chat-demo","shard":"1","boards":{"new":190,"debut":87}},{"full_name":"deepseek-ai/dsh-superagent-prototype","name":"dsh-superagent-prototype","shard":"1","boards":{"daily":114,"weekly":79,"monthly":16,"weekly_rate":133,"monthly_rate":2,"daily_forks":121,"new":15}},{"full_name":"deepseek-ai/open-infra-index","name":"open-infra-index","shard":"1","boards":{}}],"issues":[],"discussions":[{"number":275,"title":"Visible-device assignment can select ids past the last GPU when offset is nonzero","comments":0,"category":"General","created_at":"2026-11-05T08:12:29Z"},{"number":270,"title":"Profiling manager fails to start when route-based address lookup fails","comments":0,"category":"General","created_at":"2026-11-04T22:29:43Z"},{"number":268,"title":"Visible-device helpers count entries after CUDA -1 sentinel","comments":0,"category":"General","created_at":"2026-11-04T20:06:36Z"},{"number":238,"title":"Is NVIDIA H200 (Hopper, SM90) officially supported or planned?","comments":3,"category":"Q&A","created_at":"2026-10-26T12:39:20Z"},{"number":231,"title":"Roadmap for official Docker images and production serving recipes/cookbooks (similar to vLLM/SGLang)?","comments":1,"category":"General","created_at":"2026-10-26T01:31:52Z"},{"number":225,"title":"Proposal: per-request prompt mask that keeps KV but zeroes attention","comments":0,"category":"Ideas","created_at":"2026-10-25T01:43:35Z"},{"number":224,"title":"Low-concurrency decode: EP-only MoE leaves a ~7 ms/step floor (TP path for small batches?)","comments":0,"category":"Ideas","created_at":"2026-10-24T22:24:38Z"},{"number":223,"title":"CUDA_ERROR_ILLEGAL_ADDRESS from cuModuleLoadData (tilelang) at first forward — GB200 Grace ARM, single node without NCCL GIN","comments":2,"category":"Q&A","created_at":"2026-10-24T17:19:55Z"}],"latest_release":null,"readme":true,"readme_zh":false,"news":[{"title":"DeepInfer: High-performance inference framework for DeepSeek models","url":"https://github.com/deepseek-ai/DeepInfer","hn":"https://news.ycombinator.com/item?id=50163277","source":"Hacker News","lang":"en","date":"2026-10-20","points":2355,"comments":892},{"title":"What did DeepSeek release? The scope of the engine as read from GitHub","url":"https://note.com/consuloji_ch/n/n29df1c22fa0b?hl=en","source":"note","lang":"en","date":"2026-11-05"},{"title":"DeepSeek 開源 DeepInfer：正式服務用的推論引擎首度公開，同時支援 Blackwell 與昇騰 950","url":"https://www.soft4fun.net/tech/ai/deepseek-deepinfer-open-source-inference-framework.htm","source":"硬是要學SOFT4FUN","lang":"zh","date":"2026-10-22"},{"title":"Deepinfer goes open source as DeepSeek pushes cheaper AI inference","url":"https://www.mogazmasr.com/151394","source":"mogazmasr","lang":"en","date":"2026-10-21"},{"title":"DeepSeek publishes DeepInfer code for large-scale AI inference","url":"https://www.msn.com/en-ae/technology/tech-companies/deepseek-publishes-deepinfer-code-for-large-scale-ai-inference/ar-AA2fMMW2","source":"MSN","lang":"en","date":"2026-10-21"},{"title":"DeepSeek Open-Sources DeepInfer, the Inference Framework Behind DeepSeek-V4.1-Flash","url":"https://www.marktechpost.com/2026/10/20/deepseek-deepinfer-open-source-inference-framework/","source":"MarkTechPost","lang":"en","date":"2026-10-20"},{"title":"DeepSeek open-sources DeepInfer, the engine that runs its AI models","url":"https://thenextweb.com/news/deepseek-deepinfer-open-source-inference-engine","source":"The Next Web","lang":"en","date":"2026-10-20"},{"title":"DeepSeek Dials Up the Pressure on OpenAI and Anthropic With Open-Source Inference Tool","url":"https://gizmodo.com/deepseek-dials-up-the-pressure-on-openai-and-anthropic-with-open-source-inference-tool-2000828261","source":"Gizmodo","lang":"en","date":"2026-10-20"},{"title":"DeepSeek open-sources DeepInfer, the inference framework behind its production services","url":"https://siliconangle.com/2026/10/20/deepseek-open-sources-deepinfer-the-inference-framework-behind-its-production-services/","source":"SiliconANGLE","lang":"en","date":"2026-10-20"},{"title":"DeepSeek open-sources DeepInfer to cut the cost of running AI models","url":"https://cryptobriefing.com/deepseek-open-sources-deepinfer-ai-costs/","source":"Crypto Briefing","lang":"en","date":"2026-10-20"},{"title":"DeepSeek 開源推論框架 DeepInfer：支援 Nvidia、華為昇騰，但只含離線推論","url":"https://www.blocktempo.com/deepseek-open-sources-deepinfer-inference-framework/","source":"blocktempo","lang":"zh","date":"2026-10-20"},{"title":"DeepSeek 又一开源重器！深挖 DeepInfer 究竟藏着多少技术狠活？","url":"https://news.google.com/rss/articles/CBMiZ0FVX3lxTE9FTUJhTUcwa01naV9LMUxCNERHQlVOSWZMSmd2eVNaMG52MHh6Ym1vN19wV0Vobm9Hek1hV0JJcEhIaU5CN0t0Y2ZtRVJQcUV4SFViakRmN2RhRHBlOXlNbEJSR2dnVXc?oc=5","source":"雷峰网","lang":"zh","date":"2026-10-29"},{"title":"vLLM 称已追平 DeepSeek 自研引擎 DeepInfer 的性能曲线：decode 吞吐差距 6% 以内","url":"https://news.google.com/rss/articles/CBMiSEFVX3lxTE9KNW14YUFOSFFjXzBndFdmeVVMQzlkemx3TEtyWGhuZ0tvaVlXQklBa3AwanpJZ0dUdEVBMnVENE91VHplSE1IMA?oc=5","source":"智源社区","lang":"zh","date":"2026-10-25"},{"title":"DeepSeek has released 'DeepInfer,' an open-source version of its AI inference system, which is actually used in DeepSeek's production service.","url":"https://news.google.com/rss/articles/CBMid0FVX3lxTE9SZDFmUkJoMU1QMUI4ZU1qM0hkNUFyd0VMenR3TjZGMDRoVW1mc0ZxUVNZbV9yWU95cWs0TjBOR2lHYXd2WXdlMjRMX1dvdHhRM09VUThPVk83bENNWFNFdVU1NTd5RlV5SUwtdURXS21yVkhVQzln?oc=5","source":"GIGAZINE","lang":"en","date":"2026-10-21"},{"title":"DeepSeek 开源最新推理框架：单集群日吞吐 155 万亿 Token，KVCache 可保存 7 天","url":"https://news.google.com/rss/articles/CBMiWkFVX3lxTE5EcmllZ1dWbl9xWkt3b3dlbG1KR2xPVnMzcDREQ19YSTRNZWdIRXpIdk5sSTNPeVZvenFuRDcwbTMzLXZIYnJLWjJLaklTR2x1LXViT0pkMVRCQQ?oc=5","source":"InfoQ.cn","lang":"zh","date":"2026-10-21"},{"title":"刚刚，DeepSeek开源自家推理框架！","url":"https://news.google.com/rss/articles/CBMiSEFVX3lxTFBVN2kwYlRFbkVCV05zTTduYUdpcHZkWW1ONGl6ZmRSc2VydHkzVUcyUDNFLXBRTXU2T2phSzJBUWx1WU42R0NGeg?oc=5","source":"智源社区","lang":"zh","date":"2026-10-21"},{"title":"DeepSeek is open-sourcing the software it uses to run its AI models at scale","url":"https://news.google.com/rss/articles/CBMif0FVX3lxTE1IRHJQZVBMMTdReU1BbHdBX3JfNnZ4MUxfUEhZQnhzcW9zbklPa0xZZmxHWlMxeUg5QXExYld5NS02ZU1iWUlMeFdadjVpTFk2eWxLS00yR3lUMm9obl9jbE43RF8wQjZjWVlQSWpuN2J0eWN2dFdxckJDWU9ISDQ?oc=5","source":"Quartz","lang":"en","date":"2026-10-20"},{"title":"DeepInfer Explained: DeepSeek’s Open Inference Engine, Benchmarks, Specs, and Costs","url":"https://news.google.com/rss/articles/CBMiakFVX3lxTE5rbkxrbG5heUFhZlhWakp1TWdZVUVBU2xjNHB1cERYaXA1M0VTVEpfZW5HVC1ZZENIMmkzVmtOMHE0SFY1Slk0VFgwRU9xeXk5cEE3RVloX3hhV1ZVenJtRTNpR1oxRUtqTEE?oc=5","source":"Kingy AI","lang":"en","date":"2026-10-20"},{"title":"突发，DeepSeek重磅开源核心推理引擎","url":"https://news.google.com/rss/articles/CBMiTkFVX3lxTFBpUmo0VEhPblpiZDhtblBRNHJTalZSQzFKQVFiUlJ3YWFCY1RfUTUtdnhCNS00SlJxZlpyRTl6Uk9lNVhFOG13cWVGY29HQQ?oc=5","source":"36氪","lang":"zh","date":"2026-10-20"},{"title":"DeepSeek 开源推理框架DeepInfer：服务链路的最后一块拼图","url":"https://news.google.com/rss/articles/CBMiS0FVX3lxTE1HS25aQ2FVNVhFTW00d1VmV1RVNmllT0lfN0lvckUwSERLUUlPNzJxNTc5NkNmRk5oTDJHWnBtMXNUeWNETVRVWl9KYw?oc=5","source":"OSCHINA","lang":"zh","date":"2026-10-20"},{"title":"刚刚，DeepSeek又掀桌了！开源推理引擎","url":"https://news.google.com/rss/articles/CBMiVkFVX3lxTE5WZS1CT1p1RF9lQ1RlMTBuUWlvMkI3bjhfNy1zY1dLQ1ZBc1RIR3UxZGFvcXA5bUZ2REplTkJLbUNlSDdkbjNkak5PSEd5eGFPQXFaVnNn?oc=5","source":"QQ News","lang":"zh","date":"2026-10-20"}],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}