How Rankings Work

How we score

Prominence (1–100) — The primary score on every leaderboard. Prominence combines how often a product appears across query runs with how highly it is placed when mentioned. We use a modified Dowdall count: each appearance earns points inversely related to list position, so being named first counts more than being named fifth.

Rank	Dowdall	Prominence
1	1.0	100
2	0.71	90
3	0.58	85
4	0.5	81
5	0.45	78

Example assumes the product appears at that list position in every run of the scope. Actual prominence sums contributions across all runs, where a product may rank differently from one query to the next.

Those points are aggregated across all runs and models, then mapped to a 1–100 score on a log scale, so top ranks stay closer together than a straight percentage would suggest. Leaderboards are ordered by prominence; lower average position breaks ties.

Visibility % — The percentage of query runs (across all tested models) in which a product was mentioned at all. Each category is tested with 10 queries per model across 3 models, producing 30 total query events. A product mentioned in 24 of those events has 80% visibility.

Avg Position — When a product is mentioned, the average position it appears in the list of recommendations. Position 1 means the product was named first. Lower numbers are better.

Prominence, visibility, and average position are related but not interchangeable. A product can appear often at lower positions (high visibility, moderate prominence) or less often but near the top (lower visibility, strong prominence). Hover any prominence value on a leaderboard for the underlying average position and visibility.

Rank — A product's place on the category leaderboard (#1, #2, and so on), determined by prominence.

Change indicator — Movement in rank compared to the previous monthly snapshot. An upward arrow indicates improvement; a downward arrow indicates decline; a star marks products that did not appear in the previous snapshot's ranked list.

How we run queries

Each category is queried 10 times per model. Queries are framed as a developer evaluating tools — the way a real person would actually ask. For example, rather than asking "list email APIs," we ask the kind of nuanced, context-rich questions developers ask when they're genuinely choosing between options.

We do not publish the exact queries to prevent result manipulation. Queries are kept consistent across snapshots to ensure comparability. Temperature and other model parameters are held constant across all runs.

What this doesn't measure

• This is not a market share ranking. A product can have low market share but high AI prominence, or vice versa.
• This is not a quality or recommendation endorsement by LLM Rank or EveryDeveloper. Rankings reflect AI model behavior, not our editorial opinion.
• A product not appearing does not mean it was evaluated and found lacking — LLM outputs vary, and the universe of products considered by any model is not exhaustive.
• Rankings reflect AI model behavior at the time of testing. Models change, and results may differ in future snapshots.

Models tested

The current rankings are based on three models run in parallel: GPT-6.1 (OpenAI), Claude Opus 5.5 (Anthropic), Gemini 3.1 (Google). Each category leaderboard shows an aggregate prominence score and per-model icons whose ring fill reflects that model's prominence for the product. Hover a prominence value or model icon for average position and visibility. The top-5-per-model section shows each model's leading picks with model-specific prominence.

Rankings are re-run monthly. When model versions change between snapshots, that is noted in the ranking updates.

Categories

The current version covers 33 developer tool categories: AI Coding, Analytics DB, API Development, Application DBs, CDN, CI/CD, Cloud Compute, Code Hosting, Containerization, Content Management, CSS Frameworks, Data Pipeline, Documentation, Experimentation, Geo Discovery, Geocoding, Identity & Auth, Live Data, Managed Hosting, Mapping, Media Processing, Memory DB, Messaging, Observability, Open Finance, Payments, Search, Security Monitoring, Site Generators, Storage, Transactional Email, Tunneling, Vector DB .

Categories were selected based on frequency of developer decision-making and AI query relevance — these are the questions developers actually ask AI models when choosing tools.

We plan to expand to additional developer categories and other software verticals. If there's a category you'd like to see, let us know.

Ranking updates

We publish a new snapshot monthly. Each update records the models tested, any changes to categories or methodology, and what moved. See the full ranking update history →