[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:semantic-router":3},"\u003Cdiv align=\"center\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fvllm-project\u002Fsemantic-router\u002FHEAD\u002Fwebsite\u002Fstatic\u002Fimg\u002Fartworks\u002Fvllm-sr-logo.dark.png\" alt=\"vLLM Semantic Router\" width=\"50%\" \u002F>\u003Cp>An open, programmable \u003Cstrong>decision layer\u003C\u002Fstrong> for models and compute.\u003C\u002Fp>\u003Cp>\n  \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\" rel=\"nofollow ugc noopener\">Documentation\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Fapp.vllm-sr.ai\u002Fplayground\" rel=\"nofollow ugc noopener\">Playground\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002F\" rel=\"nofollow ugc noopener\">Blog\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fpublications\u002F\" rel=\"nofollow ugc noopener\">Publications\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fvllm-sr\" rel=\"nofollow ugc noopener\">Hugging Face\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Fvllm-dev.slack.com\u002Farchives\u002FC09CTGF8KCN\" rel=\"nofollow ugc noopener\">Slack\u003C\u002Fa>\n\u003C\u002Fp>\u003Ca href=\"https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F15581?utm_source=trendshift-badge&utm_medium=badge&utm_campaign=badge-trendshift-15581\" rel=\"nofollow ugc noopener\">\n  \u003Cimg src=\"https:\u002F\u002Ftrendshift.io\u002Fapi\u002Fbadge\u002Ftrendshift\u002Frepositories\u002F15581\u002Fdaily?language=Go\" alt=\"vllm-project%2Fsemantic-router | Trendshift\" width=\"250\" height=\"55\" \u002F>\n\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fvllm-sr\u002Fdecision-20\" rel=\"nofollow ugc noopener\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fvllm-project\u002Fsemantic-router\u002FHEAD\u002Fwebsite\u002Fstatic\u002Fimg\u002Fhf-trending.svg\" alt=\"Decision 2.0 — #1 on Hugging Face Trending Collections, October 6, 2026\" width=\"300\" height=\"55\" \u002F>\n\u003C\u002Fa>\u003Cp>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fvllm-project\u002Fsemantic-router\u002Factions\u002Fworkflows\u002Fmain.yml\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fvllm-project\u002Fsemantic-router\u002Factions\u002Fworkflows\u002Fmain.yml\u002Fbadge.svg\" alt=\"Main\" \u002F>\u003C\u002Fa>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fvllm-project\u002Fsemantic-router?sort=semver\" alt=\"GitHub Release\" \u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGo-1.25-00ADD8?logo=go&logoColor=white\" alt=\"Go\" \u002F>\n\u003Ca href=\"https:\u002F\u002Fdeepwiki.com\u002Fvllm-project\u002Fsemantic-router\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAsk-DeepWiki-6E56CF\" alt=\"Ask DeepWiki\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003C\u002Fdiv>\u003Chr \u002F>\n\u003Ch2>About\u003C\u002Fh2>\n\u003Cp>\u003Cstrong>Intelligence beyond any one model.\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cp>Give your agent harness one API for many models. vLLM Semantic Router selects or combines models for each call, guided by your policy.\u003C\u002Fp>\n\u003Cp>Your harness keeps the agent loop, tools, and task state. The Router chooses among configured backends across local, private, and cloud compute.\u003C\u002Fp>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Dimension\u003C\u002Fth>\n\u003Cth>Fragmented today\u003C\u002Fth>\n\u003Cth>With vLLM SR\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>\u003Cstrong>Models\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>Different models excel at different tasks.\u003C\u002Ftd>\n\u003Ctd>Select or combine models.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Compute\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>Hardware varies in speed and capacity.\u003C\u002Ftd>\n\u003Ctd>Choose among configured backends.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Location\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>Edge, private, and cloud.\u003C\u002Ftd>\n\u003Ctd>Keep calls within approved locations.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Preference\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>Priorities change by task.\u003C\u002Ftd>\n\u003Ctd>Set quality, latency, and cost priorities.\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fdocs\u002Fintro\u002F\" rel=\"nofollow ugc noopener\">Explore how it works →\u003C\u002Fa>\u003C\u002Fp>\n\u003Ch2>Getting Started\u003C\u002Fh2>\n\u003Ch3>Install\u003C\u002Fh3>\n\u003Cpre>\u003Ccode class=\"language-bash\">curl -fsSL https:\u002F\u002Fvllm-sr.ai\u002Finstall.sh | bash -s -- --channel stable\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cp>For pip, uv, or agent-driven installation, see the \u003Cstrong>\u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fdocs\u002Finstallation\u002F\" rel=\"nofollow ugc noopener\">Installation Guide\u003C\u002Fa>\u003C\u002Fstrong>.\u003C\u002Fp>\n\u003Ch3>Connect your agent harness\u003C\u002Fh3>\n\u003Cp>Point your harness at the Router's inference endpoint. Use a public model ID such as \u003Ccode>vllm-sr\u002Fauto\u003C\u002Fcode>.\u003C\u002Fp>\n\u003Cp>Follow \u003Cstrong>\u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fdocs\u002Finstallation\u002Fagent-harness\u002F\" rel=\"nofollow ugc noopener\">Connect an agent harness\u003C\u002Fa>\u003C\u002Fstrong> for setup and compatibility.\u003C\u002Fp>\n\u003Ch3>Online playground\u003C\u002Fh3>\n\u003Cp>Try the online playground at \u003Ca href=\"https:\u002F\u002Fapp.vllm-sr.ai\u002Fplayground\" rel=\"nofollow ugc noopener\">https:\u002F\u002Fapp.vllm-sr.ai\u002Fplayground\u003C\u002Fa>.\u003C\u002Fp>\n\u003Cp>Credentials:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Username: \u003Ccode>love@vllm-sr.ai\u003C\u002Fcode>\u003C\u002Fli>\n\u003Cli>Password: \u003Ccode>vllm-sr-read\u003C\u002Fcode>\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Latest News\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>[2026\u002F10\u002F06] \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fvllm-sr\u002Fdecision-20\" rel=\"nofollow ugc noopener\">Decision 2.0\u003C\u002Fa> reached #1 on Hugging Face Trending Collections.\u003C\u002Fli>\n\u003Cli>[2026\u002F10\u002F06] \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002Fvela-2-0-open-foundation-routing-models\u002F\" rel=\"nofollow ugc noopener\">Vela 2.0: Towards Open Foundation Routing Models\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>[2026\u002F09\u002F24] \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002Fv0.4-vllm-sr-hermes-release\u002F\" rel=\"nofollow ugc noopener\">vLLM Semantic Router v0.4 Hermes: Many Models, One Improving System\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>[2026\u002F09\u002F22] \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002Fdecision-models\u002F\" rel=\"nofollow ugc noopener\">Introducing Decision 1.0: Open Decision Foundation Models\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>[2026\u002F09\u002F18] \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002Fvela-models\u002F\" rel=\"nofollow ugc noopener\">Introducing Vela 1.0\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>[2026\u002F08\u002F24] \u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002Fjoin-vllm-sr-workgroups\u002F\" rel=\"nofollow ugc noopener\">Find Your Focus: How to Join and Work Together\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>[2026\u002F08\u002F05] [LettuceDetect v2 in Semantic Router: Generative Hallucination Detection as a vLLM Endpoint](\u003Ca href=\"https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002Flettucedetect-v2-generative-hallucinat\" rel=\"nofollow ugc noopener\">https:\u002F\u002Fvllm-sr.ai\u002Fblog\u002Flettucedetect-v2-generative-hallucinat\u003C\u002Fa>\u003C\u002Fli>\n\u003C\u002Ful>\n",1791678840121]