diff --git a/README.md b/README.md index 823fbb2..917c7ac 100644 --- a/README.md +++ b/README.md @@ -1,4 +1,4 @@ -# Luxe +
Themes
-- Luxe's own palette is warm graphite with a champagne-brass accent, and its hues come from a - set of paint colors rather than a terminal convention. Those are mid-luminance by nature, - so the hues were kept and the lightness re-derived against a contrast floor. -
-- Three rules do the work, and each one is asserted in the test suite rather than eyeballed: - every background and gray shares one neutral hue, every accent sits on one lightness - plateau so none of them shouts, and nothing drops below 4.5:1 on its own background. -
-{note}
-- Ports that stay true to their originals. Luxe only steps in where a color is too faint to - read a word in, or where a theme's "red" isn't red, since a failed test shown in green - says the opposite of what happened. -
-
- Point at one to see it. That's how /theme works in the terminal too: the
- cursor moves and the whole app repaints, because a palette isn't something you can judge
- from its name.
-
- Set one with "theme": "nord" in ~/.config/luxe/config.json, or
- override individual tokens with a colors map. On Omarchy, Luxe wears your
- desktop's theme and changes with it when you switch.
-
Tools
{primary.commit}, the codebase
the harness eval runs in, with its virtualenv in place. The shell row is the command's
raw output; tightest cut is the least any other agent sends its model of it,
- which is Codex's 10 KB, head and tail kept. Pi and OpenCode keep up to 50 KB and Claude
- Code 30,000 characters. A token is about four bytes.
+ which is Codex's 10 KB, head and tail kept. OpenCode keeps up to 50 KB and Claude Code
+ 30,000 characters. A token is about four bytes.
Those are single calls, and across a whole task the bars shrink, because most calls are small to begin with. What a whole task costs is below.
- Luxe, Claude Code, Codex, OpenCode and Pi each ran the first twelve Django tasks of + Luxe, Claude Code, Codex and OpenCode each ran the first twelve Django tasks of SWE-bench Verified Mini, with the same model, the same prompt and the same keys, every call metered at the network and priced from one list. Each figure is what the other agent spent on the same task, as a multiple of what Luxe spent: above 1× costs more than Luxe.
| Model | Claude Code or Codex | OpenCode | -Pi | {multiple(vendorAgent(row))} | {multiple(row.opencode)} | -{multiple(row.pi)} | )) }
|---|
Every run is charged as a cold start, the way a session begins after a break. Run side
by side on a warm cache, Claude Code and OpenCode cost up to 2.6× what Luxe does on
- Claude models. Pass rates overlapped on every model. Pi is the cheapest by design: four
- tools and a short prompt, with web search, MCP and the rest left to extensions. Graded
- on one machine rather than in SWE-bench's own containers, so these are not leaderboard
+ Claude models. Pass rates overlapped on every model. Graded on one machine rather than in SWE-bench's own containers, so these are not leaderboard
scores. Measured at {agentCost.commit}.