How we check

Every fact on a tool’s page comes from the maker’s own pages: the official site, pricing page, help center or terms. Next to each fact we show the date we confirmed it and a link to the source.

Each fact is in one of four states. Checked: confirmed on an official page, with the date. Not verified: we could not confirm it, and we say why. Does not apply: the fact makes no sense for this tool. Needs re-checking: we confirmed it more than 90 days ago, so it may be out of date.

Only checked facts count towards badges, filters and totals such as “Free to start”.

Prices and free plans are for that one kind of tool (the sub-category the card is listed under), in US dollars as shown on the maker’s US pages. A free plan for the whole product does not mean this kind of tool is free.

For commercial use we summarise the maker’s terms and quote the relevant line. That is not legal advice, and it says nothing about who owns a particular image.

“Get” and “Official site” always go to the maker’s own site. We add no tracking to these links.

How we order tools

Both lists use the 30 complete UTC days shown on the homepage. Each is sorted by its own metric. Equal readings share a rank.

Search interest covers 58/69 eligible products. Google Trends data is aligned to a common scale using a shared reference and averaged globally over the period. The topic or search term may cover the whole brand or product, including uses beyond the Agent capability shown. The index is relative and does not give search counts. Queries with nonzero readings on fewer than half the days are treated as missing.

OpenRouter usage covers 14/69 eligible products. The totals sum the 30 daily token counts attributed to each public app profile, across models. Coverage depends on app attribution and routing. Tokenizers vary by provider, and the counts do not track unique users, tasks or usage on other platforms.

Products without usable data for a list have no position in that list. Matching GitHub and website ranking data was unavailable for this period.

Google Trends · OpenRouter

Products missing data

Search interest (11) · Bubble, Cue AI, Kiro, Magnific, Make.com, Microsoft Copilot Studio, Oh My Pi, Otter.ai, Qwen Chat, Softr, Stack AI

OpenRouter usage (55) · Adobe Firefly, Base44, Bolt.new, Botpress, Browser Use, Bubble, Canva, ChatGPT, Claude, Coze, Cue AI, DeepSeek, DeerFlow, Dify, Dreamina, Durable, ElevenLabs, Framer, Gamma, Gemini, Gemini Notebook, Genspark, GitHub Copilot, Grammarly, Grok, Gumloop, HeyGen, Hostinger AI Builder, Kimi, Kiro, Krea, Langflow, Lovable, Lovart, Magnific, Make.com, Manus, Meta AI, Microsoft Copilot, Microsoft Copilot Studio, OpenCode, Otter.ai, Perplexity, Qwen Chat, Relevance AI, Replit, Runway, Softr, Stack AI, TRAE, v0, Voiceflow, Wix, Zapier, 豆包

Search interest

OpenRouter usage

Tools are ordered by how much people search for each brand on Google, worldwide: the average over the last 28 days, adjusted for whether interest is rising or falling. The data runs to 2026-11-11.

That shows how popular a brand is, not how good its images are. For tools like ChatGPT most of that interest comes from chat, not from images.

Tools we have no search data for come last, sorted by name, and the list says why.

Coding tools

Prices and free plans are for the job on that page. A free chat plan counts only if it can be used to write code, and a tool that is free but needs you to pay an AI model maker yourself is not counted as free. One-time trials are marked as such; offers with an end date stop counting once they end.

For writing code and building agents we say whether the maker trains its AI on your code or data by default, which plan and setup that applies to, whose AI model gets your data, and how to turn it off. For website builders we say whether you can take the code with you.

“Get” and install-guide links only go to addresses the maker owns: its official site, or other addresses we have shown belong to it. We do not show install commands to copy.

Coding tools are ordered the same way, each kind of tool on its own (data to 2026-11-13).

A list can use both Google Trends topics and search terms. Topics group searches for the same concept across languages. Search terms match the words we enter. We sort by the resulting scores, so bear in mind that nearby ranks may cover different sets of searches.

General assistants such as ChatGPT are ranked by interest in the whole brand, which covers every use, not only writing code.