role-model

model routing protocol and runtime, and a package for Pi by @trydotworks

[role-model @rolemodeldev]

@karimatiyeh you might want to try the role-model router https://t.co/QmdfM5a50W

[role-model @rolemodeldev]

role-model v0.0.7 now supports Kimi K3 via Kimi Code subscriptions. Go try it out.

[role-model @rolemodeldev]

role-model v.0.0.6 is live on GitHub with support for pre-served cache when routing across models, and more.

[role-model @rolemodeldev]

Get role-model: https://t.co/QmdfM5aCQu

[role-model @rolemodeldev]

New in role-model v0.0.5: A new UI.

Head over to GitHub and get your model router.

[role-model @rolemodeldev]

here's what's new in role-model 0.0.2:

- Role-aware routing and observability
- Benchmark scores now influence routing
- Major Pi integration upgrade
- Observe got faster and more useful

[role-model @rolemodeldev]

Here's how you set up role-model as your model router in Pi!

[role-model @rolemodeldev]

@sooh_hoos @try you can use role-model with Codex as well, and route between any models you like. All you need to do is install the role-model runtime, then configure the endpoint in Codex. So you can for example route between GPT and DeepSeek or GLM. Or use local models in Codex.

[role-model @rolemodeldev]

we just published the first alpha release of the role-model runtime and the pi-role-model package for Pi.

With the ‘pi-role-model’ package, Pi gains the ability to route between models and providers based on task.

[role-model @rolemodeldev]

Install for Pi
’pi install npm:@try-works/pi-role-model’

Then let your agent install and configure the role-model runtime or point it to:

‘https://t.co/QmdfM5a50W’

[role-model @rolemodeldev]

With role-model, you can set up a pool of models to route between, benchmark them, and choose routing strategies like difficulty, cost, speed or controller-based routing based on your work. Easily switch routing strategies inside Pi, and monitor routing performance.