jev-model-routing

1# Model routing with Jev
2 
3Jev reads a turn and answers three questions in one ~0.4 s request: how hard is it, what kind of work is it, and would a mistake be costly. Code then walks your pool for that tier and specialty and takes the first model that fits (images, context size). You do not pick models by feel; you ask.
4 
5## On Hermes it is automatic
6 
7With the `hermes-jev` plugin enabled, each fresh user turn is routed once, before the first model call. Tool-loop follow-ups reuse that decision only while provider, endpoint, host model and outgoing request model remain unchanged. A mismatch disables routing for the rest of that turn, including an A → B → A return, without buying another classifier answer. If the first outgoing model disagrees with host metadata, retain the outgoing request and do not classify; equivalent provider prefixes are normalized. The next turn resets the decision. Endpoint bindings stay in memory and are not written to telemetry. Switches, per profile:
8 
9```
10/jev                    status
11/jev routing shadow     decide and log, but do not switch (start here)
12/jev routing on         switch models
13/jev routing off
14/jev notice on          show "[Jev] medium · coding → kimi-k2.7-code · confidence 0.97" on routed replies
15```
16 
17A plugin can swap the model, not the provider connection. On OpenRouter that still means every vendor (DeepSeek, GLM, Kimi, MiniMax, Grok, Qwen, Gemini, GPT). If you run `/model` yourself, your choice wins and Jev stays out of the way.
18…