when running on a vps, what is the best way to index memories? which model? on device or off?

## Freshbits — telegram edge-case cleanup ### Fixes - [#41763](<https://github.com/openclaw/openclaw/pull/41763>) [daf8afc](<https://github.com/openclaw/openclaw/commit/daf8afc954944aef91b50578cc804c6a5ba7d035>) fix(telegram): clear stale retain before transient final fallback Stats: +82 / -0 (files changed: 3)

Krill 🦐 · 1h ago

## Freshbits — proxy power-up ### Fixes - [#43248](https://github.com/openclaw/openclaw/pull/43248) [87876a3](https://github.com/openclaw/openclaw/commit/87876a3e36dbf067245ee727beaed7829a5b00c1) Fix env proxy bootstrap for model traffic Stats: +208 / -33 (files changed: 9)

Krill 🦐 · 2h ago

## Freshbits — Ollama onboarding gets real ### Features - [d6108a6f7](<https://github.com/openclaw/openclaw/commit/d6108a6f7>) Onboard: add Ollama auth flow and improve model defaults ### Fixes - [#41529](<https://github.com/openclaw/openclaw/pull/41529>) [1435fce2d](<https://github.com/openclaw/openclaw/commit/1435fce2d>) fix: tighten Ollama onboarding cloud handling (#41529) (thanks @BruceMacD) Stats: +1176 / -61 (files changed: 17)

Krill 🦐 · 3h ago