blog(autorouter): setup, testing, and going beyond complexity routing

blog/autorouter_setup_and_testing/index.md

@@ -0,0 +1,100 @@
+---
+slug: auto-router-setup-and-testing
+title: "Auto-Router: Get Started in 1 Command, & Go beyond Complexity Routing"
+date: 2026-08-05T10:00:00
+authors:
+  - tin
+image: ./hero.png
+description: "Six changes to the Auto-Router: a one-line agent setup skill, Test Routing in the UI, Anthropic and OpenAI family presets, a replaceable classifier prompt, renameable tiers, and configurable reminder markers."
+keywords: [auto router, llm routing, litellm, model routing, claude code, llm classifier, complexity router]
+tags: [routing, complexity-router, ui, engineering]
+hide_table_of_contents: false
+---
+
+![LiteLLM Auto-Router: get started in 1 click, go beyond complexity routing](./hero.png)
+
+We've made it easier than ever to setup and test your Auto-Router, and with customizable tier names + classifier system prompts you can go beyond complexity routing. 
+
+{/* truncate */}
+
+:::info Availability
+
+Everything below ships in **v1.97.x**.
+
+:::
+
+:::info[🚀 Help shape the Auto-Router]
+
+Get early access, work directly with the LiteLLM team, and influence the roadmap with your production traffic.
+
+<a className="button button--primary button--lg" style={{background: '#2e8555', borderColor: '#2e8555', color: '#fff'}} href="https://calendar.app.google/i2e7qVEJphHi5S8UA">Apply to Become a Design Partner</a>
+
+<br /><br />
+
+Already testing it? Share your results in [discussion #32168](https://github.com/BerriAI/litellm/discussions/32168).
+
+:::
+
+## Let your agent set up the router
+
+Paste this into Claude Code, Codex, Cursor, or any agent with shell access:
+
+```
+run curl -fsSL https://docs.litellm.ai/skills/auto-router and follow the instructions
+```
+
+- It reads the models your proxy already serves, then interviews you for the router name and the model behind each tier
+- It writes the config for you, whether your proxy is file-based or DB-managed
+- Before finishing it lists the defaults it left in place, with what changing each one buys, and asks whether you want any changed
+
+## Test Routing in the UI during setup
+
+![Test Routing modal showing the tier, score, signals and routed model for a test prompt](./test-routing.png)
+
+- **Test Routing** now sits beside Test Connection on the Add Auto Router form
+- Send a prompt, see the tier it lands in and why, against the config currently on screen
+- Nothing is created and nothing is sent to the routed model, so it costs nothing beyond your LLM classifier (if enabled)
+
+
+## Presets for the Anthropic and OpenAI families
+
+- Add Auto Router now opens on a name field and a **Template** dropdown: Anthropic family, OpenAI family, or Custom
+- Picking a family fills in the whole config and collapses the detail behind a one-line tier summary
+- Presets that reference a model your proxy doesn't serve grey out and tell you which one is missing
+
+## Replace the classifier's system prompt
+
+The LLM classifier shipped with one built-in rubric, so the router could only grade complexity. `classifier_llm_config.system_prompt` now allows you to define your own routing criteria — whether you want a more in-depth prompt for complexity or routing based on another criterion such as data sensitivity or model capability (vision, audio, image).
+
+![Classification Method panel with the Classifier Prompt override and fallback options](./classifier-prompt.png)
+
+- `classifier_fallback` decides what happens when classification fails: the heuristic scorer, or straight to `default_model`
+
+## Rename the tiers
+
+Along with the above change, you can now change the tier names from the default: SIMPLE / MEDIUM / COMPLEX / REASONING. If your team prefers Fast / Standard / Premium / Deep or Image / Video / Audio / Text, an optional `tier_labels` map renames them.
+
+- Names change in the dashboard, the spend logs, and the LLM classifier's rubric, so the classifier reasons in your vocabulary
+- Display-only. Config keys stay canonical, routing behavior doesn't move, and API callers never see these names
+- Partial maps are fine; unlisted tiers keep the default name
+
+## Configurable reminder markers, easy OpenClaw integration
+
+The router strips harness-injected context before classifying, so a token-budget note doesn't get graded as the user's actual question. That marker pair was hardcoded to `<system-reminder>`; a new `reminder_markers` field lets harnesses like OpenClaw use their own markers.
+
+```yaml
+complexity_router_config:
+  reminder_markers:
+    - "<<<begin_ctx>>>"
+    - "<<<end_ctx>>>"
+```
+
+## Try it
+
+:::info
+
+Start with the one-line agent command, or open Add Model → Auto Router in the dashboard and pick a family preset. Questions and results in [discussion #32168](https://github.com/BerriAI/litellm/discussions/32168), or [apply to be a design partner](https://calendar.app.google/i2e7qVEJphHi5S8UA) to work on this with us directly.
+
+:::
+
+Full reference on the [Auto Routing docs page](/docs/proxy/auto_routing).