vllm-project /semantic-router

An open, programmable decision layer for models and compute.

★ 6,479⑂ 1,122👁 61⑃ 3,930● 2,693▤ 4
GoApache-2.0Created 2025-08-26vllm-sr.ai
More from this author
Download: Release v0.4.0Release-0.4.0 ↓0.4.02026-09-27
+2
Daily
6,479
stars
1,122
forks
61
watchers
3,930
prs
2,693
issues
4
releases

README

An open, programmable decision layer for models and compute.


  Documentation |
  System Duo |
  Blog |
  Publications |
  Hugging Face |
  Slack



  


  



!GitHub Release
!Go




---

 About

Intelligence beyond any one model.

Give your agent harness one API for many models. vLLM Semantic Router selects or combines models for each call, guided by your policy.

Your harness keeps the agent loop, tools, and task state. The Router chooses among configured backends across local, private, and cloud compute.

| Dimension | Fragmented today | With vLLM SR |
| --- | --- | --- |
| Models | Different models excel at different tasks. | Select or combine models. |
| Compute | Hardware varies in speed and capacity. | Choose among configured backends. |
| Location | Edge, private, and cloud. | Keep calls within approved locations. |
| Preference | Priorities change by task. | Set quality, latency, and cost priorities. |

Explore how it works →

 Getting Started

 Install

bash
curl -fsSL https://vllm-sr.ai/install.sh | bash -s -- --channel stable


For pip, uv, or agent-driven installation, see the Installation Guide.

 Try it online

Try System Duo: pick an example and it runs live in your browser, with a free key on first run.

 Latest News

- [2026/10/14] d3 ranks 1 on text and vision in the Jev Decision Index, with smaller sizes on the Pareto frontier.
- [2026/10/14] Decision 3.0, Decision 2.0, and Vela 2.0 reached Hugging Face Trending Collections.
- [2026/10/06] Vela 2.0: Towards Open Foundation Routing Models
- [2026/09/24] vLLM Semantic Router v0.4 Hermes: Many Models, One Improving System
- [2026/09/22] Introducing Decision 1.0: Open Decision Foundation Models
- [2026/09/18] Introducing Vela 1.0
- [2026/08/24] Find Your Focus: How to Join and Work Together
- [2026/08/05] LettuceDetect v2 in Semantic Router: Generative Hallucination Detection as a vLLM Endpoint


Earlier announcements

- [2026/07/21] Beyond a Single Model: Building Mixture-of-Models Systems with vLLM Semantic Router
- [2026/07/09] Adding Cursor-Style Auto Model Selection to OpenCode with vLLM Semantic Router
- [2026/06/29] Micro-Agent: Beat Frontier Models with Collaboration inside Model API
- [2026/06/16] Beyond One Model: Fusion in vLLM Semantic Router
- [2026/06/05] vLLM Semantic Router v0.3 Themis: From Signals to Stateful Production Routing
- [2026/03/24] Vision Paper Released: The Workload-Router-Pool Architecture for LLM Inference Optimization
- [2026/03/10] v0.2 Released: vLLM Semantic Router v0.2 Athena Release
- [2026/02/27] White Paper Released: Signal Driven Decision Routing for Mixture-of-Modality Models
- [2026/01/05] Iris v0.1 Released: vLLM Semantic Router v0.1 Iris: The First Major Release
- [2025/12/16] Collaboration: AMD × vLLM Semantic Router: Building the System Intelligence Together
- [2025/12/15] New Blog: Token-Level Truth: Real-Time Hallucination Detection for Production LLMs
- [2025/11/19] New Blog: Signal-Decision Driven Architecture: Reshaping Semanti

Top issues

Discussions

Latest release

Release v0.4.0 v0.4.0 2026-09-27
Announcement

982 commits · 130 contributors · 107 first-time contributors since Themis. This is a milestone built by a growing community across models, runtime, evaluation, and the experience of putting it all to work.

Read the full release blog 🎉

 Python

bash
pip install vllm-sr==0.4.0


 New Contributors
 @xiaoyu-xyz made their first contribution in https://github.com/vllm-project/semantic-router/pull/2061
 @wyf027 made their first contribution in https://github.com/vllm-project/semantic-router/pull/2187
 @omkar-334 made their first contribution in https://github.com/vllm-

Snapshot: 2026-11-16T09:23:14Z · Data license Apache-2.0 · JSON