Skip to content

v0.1.1 — Bring your own brain

Choose a tag to compare

@ThinkCode ThinkCode released this 20 Oct 03:17
· 9 commits to main since this release

v0.1.1 — Bring your own brain

The intelligence layer is now a choice, switchable with one click from the Copilot header (⇅ Switch):

  • Apple Intelligence (on-device) — macOS 26+ on Apple Silicon. No download, no account, no key: briefings, Copilot and alert explanations run on the model built into macOS, and nothing ever leaves the Mac. On Intel or pre-26 systems the option hides itself.
  • Local Ollama — free and private; ollama pull qwen3:4b is a great fit for an 8 GB Mac.
  • Any OpenAI-compatible endpoint — LM Studio, vLLM, or a cloud router. The switcher remembers your custom endpoint, so you can round-trip to local Ollama and back without retyping it.

Also in this release:

  • Copilot answers rewritten for small models — one-sentence takeaway first, at most two supporting details, no more Label: value recitals.
  • A first-run chooser appears if no model is configured: connect intelligence in one click, or skip it — everything except briefings/Copilot works without any model.

Full control (endpoint, key, model) stays in ⚙ Settings.

Checksums (SHA-256)

d03e15c66266753cc04838c123ec3072d8f54935576d7bb9b2618c06ca1ba7a1  System_Cockpit_0.1.1_aarch64.dmg
1b359b9b6446f5264ca1a581a5871725029e9ddaf09a889b582771f836bb5c1b  System_Cockpit_0.1.1_x64.dmg

Both DMGs are Developer ID signed, hardened runtime, notarized and stapled — no Gatekeeper warnings.