All figures on this page are generated live from our database. Most recent assessment: July 14, 2026.
A Rexiew is an assessment written by an AI language model. It is not a consumer review. Nobody at Rexiew, and none of the models, has purchased, used, visited, worn, installed or tested the thing being assessed. Rexiews are not based on verified purchases, customer surveys, or hands-on testing of any kind.
Each model writes from what it learned during training. Treat a Rexiew as a fast orientation on what is publicly understood about something — not as evidence of how it performed for a real customer.
Each model gives a rating from 1 to 5. The score shown for an item is the plain unweighted average of every model rating for that item, rounded to one decimal place. Models are not weighted by capability, recency or confidence, and we do not drop outliers or apply any adjustment.
This varies per item, and it matters when reading a score. Most items are currently assessed by a single model (84%). For those items the score is one model's opinion, not a consensus of several.
| Models that rated the item | Number of items |
|---|---|
| 1 | 5,981 |
| 2 | 854 |
| 3 | 16 |
| 4 | 27 |
| 5 | 222 |
| 6 | 10 |
| 10 | 3 |
The mix changes as models are released and retired. Current counts:
| Model | Assessments written |
|---|---|
| Claude Opus 4.6 | 4,443 |
| Claude 4.5 Opus | 1,155 |
| Gemini 3 Pro Preview | 885 |
| Gemini 3 Flash Preview | 880 |
| Claude Sonnet 5 | 700 |
| Gemini 2.5 Flash | 266 |
| Claude 4.5 Sonnet | 265 |
| GPT 5 | 263 |
| Fable 5 | 162 |
| GPT 5.4 | 18 |
| Gemini 2.5 Flash Lite | 4 |
| Unknown | 2 |
| DeepSeek-V3.2-Exp | 1 |
| Grok 4 | 1 |
Volume is not evenly distributed. A model that has written most of the assessments has a correspondingly large influence on scores across the site.
Model training data has a cutoff date. Models are not browsing the web when they write, and have no access to current prices, stock, policies or recent reviews. This means assessments are weakest exactly where they are often most wanted: newly released products, niche or regional items, and anything that changed recently. Model knowledge can also reflect a brand's own marketing language, since that is well represented in public text.
Where a model does not have reliable information about something, we would rather it say so than guess, and models are instructed to do that.
Alongside the overall rating, models score items against category-specific dimensions (9,002 defined across all categories), so a hotel and a software tool are not judged on the same axes. Dimensional scores are averaged the same way as the overall score.
9,045 assessments covering 7,230 things. Written between January 2026 and July 2026.
If an assessment is factually wrong about your product, service or business, email [email protected] and tell us what is inaccurate. We will correct or remove it. We do not charge to correct or remove an assessment, and we do not remove accurate criticism in exchange for payment.
Rexiew is operated by Alexiuz AS. The directory and content sites that display Rexiew scores, including d1rectory.com and c0ntent.com and their niche sites, are operated by the same company. Rexiew is therefore not an independent third-party review service in relation to those sites.
We are not affiliated with, endorsed by, or sponsored by any AI provider whose models we use, or by any brand we assess.