Skip to content

feat: improve catalog browsing on mobile and at scale#36

Open
brunobuddy wants to merge 2 commits into
mainfrom
audit/ui-ux-improvements
Open

feat: improve catalog browsing on mobile and at scale#36
brunobuddy wants to merge 2 commits into
mainfrom
audit/ui-ux-improvements

Conversation

@brunobuddy
Copy link
Copy Markdown
Member

💭 Why

A UI/UX audit found the catalog hard to use on phones (the header overflowed at 375px, the parameter table broke) and hard to scan as a flat list of 140 models. A couple of model names also rendered wrong.

✨ What changed

  • Header collapses into a menu on mobile; theme toggle stays visible
  • Parameter table scrolls sideways on narrow screens instead of crushing columns
  • Search/filter bar stays pinned while scrolling the whole list
  • Models group by provider, with a sort control (provider / name / params)
  • The 36 parameter chips collapse behind a "show all" expander
  • Added a clear-all-filters control, a search clear button, and "/" to focus search
  • aria-pressed on every filter toggle; provider-card chevrons; model rows link directly
  • Fixed labels: "Claude Opus 4 20250514" and "Gpt 4 Turbo 2024-04-09" (was "4.20250514")
  • Glossary now distinguishes "Max completion tokens" from "Max tokens"

👤 For users

Mobile browsing works: no clipped header, readable tables. Filters stay reachable while scrolling, and the parameter list isn't a wall of 36 chips.

📝 Notes

Also fixes a dist/ mkdir race so a clean npm run dev starts without an ENOENT.

Mobile:
- Collapse header nav into a menu so it no longer overflows at 375px
- Let the parameter table scroll horizontally instead of crushing columns

Browsing:
- Pin the search/filter bar while scrolling the full model list
- Group models by provider with a sort control (provider / name / params)
- Collapse the 36 parameter chips behind a "show all" expander
- Add a clear-all-filters control, a search clear button, and "/" to focus search

Fixes:
- Stop date stamps fusing onto versions (e.g. "Claude Opus 4 20250514")
- Render hyphenated dates cleanly ("Gpt 4 Turbo 2024-04-09")
- Distinguish "Max completion tokens" from "Max tokens" in the glossary

Also: aria-pressed on filter toggles, provider-card chevrons, direct links on
model rows, and a dist/ mkdir so a clean npm run dev works.
@vercel
Copy link
Copy Markdown

vercel Bot commented May 26, 2026

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
modelparams.dev Ready Ready Preview May 26, 2026 3:37pm

Request Review

@github-actions github-actions Bot added model Add a model that's missing site Website code or tooling labels May 26, 2026
Revert the per-model YAML label edits and instead normalize
max_completion_tokens to "Max completion tokens" at render time
(glossary, model pages, JSON-LD). The catalog data and JSON API
stay faithful to source, mirroring how model slugs are prettified
for display only.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

model Add a model that's missing site Website code or tooling

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant