> ## Documentation Index
> Fetch the complete documentation index at: https://comfyui-mcp.artokun.io/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# July 9: LM Studio & llama.cpp backends, phone pairing, gemma4 Arena results

> Every local runtime becomes a first-class backend (LM Studio, llama.cpp, custom OpenAI-compatible endpoints), a token-gated phone-pairing listener and render mailbox land for mobile, graph_query filters big workflows, and the gemma4 fine-tune ladder posts its Arena scores. The panel keeps working when you switch sidebar tabs.

*July 9, 2026*

v0.28.0 → v0.30.0 plus panel 0.6.9 → 0.7.3. The local-LLM story from the day before
rounds out to **every major local runtime**, the first **mobile** building blocks
land, and the [gemma4 fine-tune ladder gets honest Arena numbers](../blog/gemma4-fine-tune-arena).

## More local backends: LM Studio, llama.cpp, custom endpoints

* **LM Studio** as a first-class local backend (#160), with a full hands-off
  model/server lifecycle (#164) — the agent can load the model and start the server
  itself.
* **llama.cpp** (`llama-server`) as a first-class local backend (#161).
* **Custom OpenAI-compatible endpoint** as a first-class backend (#162) — point the
  panel at anything that speaks the API.
* Temperature pinned to 0 for these backends — nondeterministic empty finals after
  tool results are gone (#166).

Each ships with its own panel chip + setup card (panel 0.6.9 / 0.7.0).

## Mobile building blocks: phone pairing + render mailbox

* **On-demand phone pairing** (#180) — a token-gated LAN/tunnel listener; nothing is
  exposed until you ask.
* **Render mailbox** (#182) — never lose a render while the phone is away; headless
  (mobile/remote) tabs get renders delivered in-turn (#165) and `show_media` can
  inline media bytes for clients with no ComfyUI reachability (#171).

## gemma4 fine-tune Arena results

LLM Arena results for the `gemma4-comfyui-mcp` ladder: **`:e4b` 14/20** (best local
model tested), `:12b` 13/20, and `:e2b` honestly flagged at 4/20 pending its v2
training fix. Leaderboard SVGs + model-sizing guidance updated across the docs and
the Ollama onboarding copy. [Full story on the blog](../blog/gemma4-fine-tune-arena).

## Workflow intelligence + node development

* **`query_workflow`** (#179) — filter / traverse / aggregate over big workflows
  without loading the whole graph into context.
* **Graph-health findings** in `validate_workflow` / `analyze_workflow` (#175) —
  disconnected nodes, missing required inputs, duplicate model loads, orphaned
  branches, muted/bypassed nodes.
* **Node-dev tools** (#173) — path-jailed read/search/write/patch plus per-pack git
  for `custom_nodes`; commit/push stays behind `COMFYUI_MCP_ALLOW_GIT_WRITES`.
* **`get_comfyui_settings` / `set_comfyui_setting`** (#174) — read/write ComfyUI's
  own user-settings store. **`calculate`** (#176) — safe batch math with variables
  and seeded RNG.
* **`panel_auto_layout`** (#177) — one-shot topological canvas auto-arrange, and
  `panel_connect` auto-matches slots by type with full diagnostics on failure (#178).

## Panel: survives tab switching

The agent now **keeps working when you switch to another sidebar tab** (0.7.3) — the
panel detaches instead of tearing down, so the connection, session, and chat survive,
and an **activity badge** on the tab icon shows a green spinner while a turn runs and
a red dot when it finished while you weren't looking. Bridge-driven graph mutations
became properly undoable (one command = one Ctrl+Z, 0.7.1), and a "Remote control"
QR button (behind a default-off beta toggle) previews phone pairing.