Tried out @routstr @Private Provider using Pi agent. Initially I got "Response was truncated before completion. ". Afterwards, it stopped working altogether. "EHBP upstream tinfoil returned 404 for model kimi-k3".
Switched to @PayPerQ and it works flawlessly. Same private models.
We need private models to prevent leaking everything we do exclusively to Big Labs. More on this soon.
Login to reply
Replies (1)
Hey, sorry about that. Look into this right now.
It seems to be working as expected now. We'll investigate why it return 404 back then.
Also, we're adding thinking selection to Pi soon. 
GitHub
feat: expose per-model reasoning effort on /v1/models by sh1ftred · Pull Request #725 · Routstr/routstr-core
What
GET /v1/models now returns the per-model reasoning object OpenRouter already publishes (supported_efforts, default_effort, default_enabled, ma...