The Spinoza Score

The Spinoza Score is our verdict on whether an AI tool is actually any good, distilled to a single number from 0 to 5. It's the average of five things we test, scored the same way across an entire category — so a 4.2 means the same thing whether you're looking at an image model or a coding agent.

No vendor claims. No user-survey averages. No marketing math. We test the tool, then we score it.

Every Spinoza Score is the average of five dimensions, each rated 0 to 5. The labels shift to fit the category, but they always answer the same five questions:

• How good is the output? The quality of what the tool actually produces — on identical prompts and real tasks, not cherry-picked demos.
• How much can it do? Depth and range against direct rivals. Not the feature list — what holds up when you actually use it.
• How good is the experience? How fast you reach a usable result, and how hard the tool makes you work for it.
• Can you rely on it? Consistency, uptime, and whether it performs the same next month as it does today.
• Is it worth the money? The real cost of a result once you strip out credits and marketing math — not the sticker price.

We test before we score. Flagship tools get hands-on, documented runs — the same prompts and tasks across every competitor in the category. Every number is backed by something real: actual outputs, actual pricing math, recorded conditions. If we can't defend a score, we don't publish it.

AI tools change weekly, so every review carries the date it was last tested — and we re-score the moment something material ships: a major version, a new model, a pricing change. The Score is a living verdict, not a one-time stamp.

We don't score every product. A few things keep a tool off the site, no matter how much buyer interest there is:

• No real free tier or trial — we can't score what we can't use.
• Coverage conditioned on approval — we don't show drafts before publishing, and we don't negotiate wording.
• Dead products — tools that have stopped shipping. If the velocity is zero, so is our interest.

The Spinoza Score can't be bought. No tool pays for a rating, a rank, or a revision, and the affiliate links that fund our work never move a number. It's not a popularity contest either — we don't average internet sentiment, we run the tests ourselves.

One number. The same standard for everyone. That's the whole point.

We correct errors fast and visibly. Every correction is appended to the original piece, dated, and explained. To flag one, write to kamran@thespinozajournal.com.

Want your product scored? Submit it here.