Skip to content

fix(dashboard): collapse image model scope into one opt-in row (#116)#120

Merged
teamchong merged 1 commit into
mainfrom
fix/116-collapse-model-rows
Jul 18, 2026
Merged

fix(dashboard): collapse image model scope into one opt-in row (#116)#120
teamchong merged 1 commit into
mainfrom
fix/116-collapse-model-rows

Conversation

@teamchong

Copy link
Copy Markdown
Owner

The dashboard always showed three full-width rows of model-family chips (Claude / OpenAI / Grok), even though only Fable 5 is enabled by default.
That's a lot of pixels for something almost nobody should touch.

Now it's a single collapsed line. Expanding it is an explicit opt-in.

Before:

  Image Claude models    [Fable 5 ✓] [Opus] [Haiku] …
  Image GPT models       [GPT 5.5] [GPT 5.6] …
  Image Grok models      [Grok 4] [Grok 4.5] …

After:

  ▸ Image model scope    Fable 5 only by default · expand to experiment

Expanding shows:

  • the same chips, plus a PXPIPE_MODELS CSV box (chips and box stay in sync; runtime-only, export the env var to persist)
  • a warning that compression is only tuned for Fable 5 — other families can use more tokens, not less
  • a "routing help" modal: OPENAI_MODELS / CLOUDFLARE_MODELS routing, and how to switch with /model claude-<model> (e.g. /model claude-moonshotai/kimi-k3)

When compression is off, a single "settings have no effect" note renders once instead of repeating on every row.

Fixes #116

@teamchong teamchong self-assigned this Jul 18, 2026
Three full-width rows of model chips were always visible, but only
Fable 5 is enabled by default. Now it's one collapsed line; expanding
shows chips + PXPIPE_MODELS CSV, a Fable-5-only warning, and a routing
help modal (/model claude-<model> to switch).

Fixes #116
@teamchong
teamchong force-pushed the fix/116-collapse-model-rows branch from d6aa82d to 13f737d Compare July 18, 2026 17:17
@teamchong
teamchong merged commit 2398337 into main Jul 18, 2026
1 check passed
@teamchong
teamchong deleted the fix/116-collapse-model-rows branch July 18, 2026 17:18
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

The rendered images are low-fidelity; let me pull exact source lines for the spots I need to edit.

1 participant