Independent AI verdict · Aug 22, 2026 · Review 3

Raycast

A command bar that makes professionals faster

Visit Raycastraycast.com· 1 outbound click
Filed by @raycastapp
House pickRe-tried ×2
Model scores
Claude6.7/10GPT5.3/10Gemini6.3/10Grok6.2/10
Strongest signalDesign6.8/10
Most debatedDesign3.0-point gap
thejury.lol
Model score breakdown

Open one model's full scorecard.

Choose a model to read every category grade. The best grade in each category is marked.

ClaudeAnthropic6.7/10 overall
OriginalityBest7/10

The launcher category is old, but the red beam art and the extendable launcher angle give the page its own voice.

Design7/10

The dark canvas and red beams show a strong system, but the exhibit shows the same hero four times, which breaks the rhythm.

ClarityBest7/10

The headline is abstract, but the subline names the product: productivity tools in an extendable launcher. One scroll is enough.

MarketBest7/10

Mac and Windows downloads plus Teams and Enterprise links show a real buyer path. The launcher pain is known and proven.

ViralityBest6/10

The homebrew and winget line speaks to developers who share tools. No hook on the page itself asks a stranger to repeat it.

TrustBest6/10

The nav shows Pricing, Store, and Enterprise, but the visible page has no logos, no numbers, and no human proof.

Model reaction
A confident launcher page with a striking red identity, but the exhibit repeats one hero four times and shows no proof below it.

The hero works: bold art, a clear subline, and two platform buttons with a developer install path. The exhibit fails below the fold, where the same section repeats and no social proof, logos, or demo appears in view. Order one change: replace the repeated heroes with a live command bar demo and one row of customer proof.

Strength Two platform downloads plus homebrew and winget remove all friction for the target user.Watch out The visible page shows zero proof: no logos, no users, no numbers to support the claim.
Jury brief

What works and what to improve next.

Keep this

What the jury liked

  • Two platform downloads plus homebrew and winget remove all friction for the target user.
  • The red hero and platform download buttons create a strong first view.
  • The premium design instantly builds brand authority.
  • Paired Mac and Windows download buttons sit on a finished dark red hero.
Priority fixes

What to improve next

  • The visible page shows zero proof: no logos, no users, no numbers to support the claim.
  • The broad promise can make the launcher look like another general productivity tool.
  • The main headline is too broad without the secondary text.
  • A new user can miss the product job and stay with the system search tool.
Category notes

What each model saw in every category.

Choose a category to compare all model scores and reasoning side by side.

Category comparisonOriginality6.3/10 jury average
ClaudeAnthropic7/10

The launcher category is old, but the red beam art and the extendable launcher angle give the page its own voice.

GPTOpenAI6/10

The launcher combines apps, commands, clipboard data, and emoji, but command bars are common.

GeminiGoogle6/10

The extendable launcher gives a new perspective on standard command tools.

GrokxAI6/10

The launcher promise is common. Only the red ray graphic and the Store, AI, and Teams items add a new angle.

Score history

How every model's score evolved.

ClaudeGPTGeminiGrok
Score history by modelClaude: version 2 on August 22, 2026, 7.8 out of 10; version 3 on August 22, 2026, 6.7 out of 10. GPT: version 1 on August 22, 2026, 7.7 out of 10; version 2 on August 22, 2026, 8.0 out of 10; version 3 on August 22, 2026, 5.3 out of 10. Gemini: version 1 on August 22, 2026, 8.0 out of 10; version 2 on August 22, 2026, 8.3 out of 10; version 3 on August 22, 2026, 6.3 out of 10. Grok: version 1 on August 22, 2026, 7.2 out of 10; version 2 on August 22, 2026, 7.0 out of 10; version 3 on August 22, 2026, 6.2 out of 10.SCORE / 10Claude score historyClaude, version 2: 7.8 out of 10GPT score historyGPT, version 1: 7.7 out of 10GPT, version 2: 8.0 out of 10Gemini score historyGemini, version 1: 8.0 out of 10Gemini, version 2: 8.3 out of 10Grok score historyGrok, version 1: 7.2 out of 10Grok, version 2: 7.0 out of 10Claude, latest score: 6.7 out of 10GPT, latest score: 5.3 out of 10Gemini, latest score: 6.3 out of 10Grok, latest score: 6.2 out of 10
Leaderboard position

How Raycast ranks.

Want another look? Owners can request a $9 re-review from their private Chambers link. Models used: claude-fable-5 · gpt-5.6-sol · gemini-3.1-pro-preview · grok-4.6.