consumer AI tools listed on the largest curated directory
Source: There’s An AI For That, as of July 17, 2026
Ask “how many AI tools are there” and you will get answers ranging from a few thousand to several million, all technically correct. The reason is simple: “AI tool” is not one thing. A consumer app you click, a model a developer downloads, an open-source repository, and a Python package are all “AI tools” by some definition, and every source counts only its own slice. This page lays out all six layers with a real number, a public source, and an as-of date for each, so you can quote the figure that actually matches your question.
Quick answer: there is no single number. As of July 2026, curated directories list roughly 51,000 consumer AI tools, Hugging Face hosts about 2.9 million downloadable models, GitHub carries around 219,000 repositories under its largest AI topic, and PyPI lists 10,000+ AI Python packages. These pools overlap heavily and can never be added into one total.
Quick Reference: The Short Answer
The short answer is that how many AI tools exist depends entirely on the source you count. Each figure below is one source’s own headline number, with its pull date. Ranked largest to smallest by what it measures.
51,242
2.9M+
219,362+
4K+
2.5K+
What Counts As An “AI Tool”?
The count changes wildly depending on the layer you measure. A packaged app a person clicks is one definition. A set of trained weights a developer downloads is another. These are the six layers people mix up when they quote a number, each with its own source and its own count.
| Term | What it is | Example | Rough count |
|---|---|---|---|
| Consumer AI tool | A packaged app or service an end user interacts with directly. | ChatGPT, Midjourney, Perplexity | ~51K |
| AI model | A set of trained weights you can download and run yourself. | Llama 3, Stable Diffusion, Whisper | ~2.9M |
| Open-source AI project | A public code repository tagged with an AI or ML topic. | PyTorch, LangChain, Ollama | ~219K |
| AI Python package | An installable library classified under the AI topic on PyPI. | transformers, scikit-learn, langchain | 10K+ |
| AI dataset | A training or evaluation data collection hosted for reuse. | ImageNet, The Pile, Common Crawl | ~963K |
| AI demo / Space | A hosted, runnable ML app or demo. | Hugging Face Spaces | ~1.4M |
The Five Layers, By Size
Stack the ecosystem by layer and the spread is enormous: a few tens of thousands of polished consumer apps sit on top of millions of raw models and projects. Most of what people call a “new AI tool” is a thin wrapper around a model in the layer beneath it.
Consumer SaaS tools
~51K
End-user apps, most built on top of a handful of foundation models.
There’s An AI For That
Hosted models
~2.9M
Trained checkpoints anyone can download and run.
Hugging Face
Open-source AI projects
~219K
Public repositories tagged with an AI or ML topic (largest topic).
GitHub topics
AI Python packages
10K+
Installable libraries classified as AI on PyPI.
PyPI
Datasets and demos
~963K + ~1.4M
Reusable training datasets plus hosted, runnable ML Spaces.
Hugging Face
How Many AI Models Are On Hugging Face?
Hugging Face is the largest open hub for hosted models, datasets, and Spaces. It is where the “machine-countable” AI pool is by far the biggest, dwarfing every consumer directory. Data as of July 17, 2026.
2.9M
models
Trained checkpoints anyone can download and run, from small classifiers to full language models.
963K
datasets
Training and evaluation data collections that models are built and tested on.
1.4M
Spaces
Hosted, runnable demos and apps that let anyone try a model in the browser.
AI Projects On GitHub, By Topic
GitHub tags repositories by topic. A single repo can carry several AI topics at once, so these counts overlap and must never be summed into one total. The “machine-learning” tag alone is larger than the entire consumer AI-tool market.
A repo can carry several AI topics at once, so these counts overlap and are never summed into one total.
How Fast Is The Count Growing?
Both consumer tools and hosted models have climbed steeply since 2021. Each line below is scaled to its own 2026 peak so the shape of the growth is comparable, not the absolute values.
44
new AI tools added per day across consumer directories, based on the There’s An AI For That new-tools counter on July 17, 2026. A live figure that swings day to day, not a stable long-run average.
AI Tool Facts Worth Knowing
Comparisons pulled straight from the numbers above, no hype, just what the data says when you put two figures side by side.
Hugging Face alone out-models every AI tool directory combined
Add up every tool tracked by There’s An AI For That, Futurepedia, and OpenTools and you get about 57.7K consumer AI tools. Hugging Face lists 2.9M+ trained models on its own, about 51 times more.
Hugging Face + directory counts
There are more AI datasets than consumer AI tools
There’s An AI For That lists 51,242 consumer-facing AI tools. Hugging Face alone hosts 962,657 datasets used to train and evaluate models, roughly 19 times as many.
Hugging Face
GitHub’s “machine-learning” topic alone beats the entire consumer AI-tool market
GitHub’s “machine-learning” topic tags 219,362 repositories, about 4.3 times the 51,242 consumer AI tools on There’s An AI For That. And that is one topic tag out of several tracked here.
GitHub topics
Hugging Face’s model count is growing faster than tool directories
Since 2021, Hugging Face’s model count has grown about 97x, while directory-tracked consumer AI tools grew about 51x over the same span. The model layer is compounding roughly 1.9 times faster.
Hugging Face + directory growth data
New AI tools get added daily, but do not extrapolate too far
The counter added about 44 new tools on July 17, 2026 alone. Sustained for a year that pace would add roughly 16,060 tools, but it swings day to day, so treat the yearly figure as a rough ceiling, not a forecast.
There’s An AI For That daily counter
Why The “How Many AI Tools” Number Keeps Moving
Every “how many AI tools exist” figure is a snapshot of one definition, taken on one day. There’s An AI For That counted 51,242 consumer-facing tools as of July 17, 2026, a number that was already out of date by the time you read this, because directories like this one add new listings daily and quietly retire dead ones without announcing it.
The growth curve behind that number is not gentle. Hugging Face’s hosted model count went from 30,000 in 2021 to 2.9M today, roughly 97x growth in 5 years, with 1,200,000 models already logged by 2024. Directory-tracked consumer tools grew too, from 1,000 in 2021 to 51,242 now, about 51x, real growth but slower than the model layer underneath it. That gap matters: most of what gets called a “new AI tool” today is a thin wrapper around an existing model, not a new model itself.
GitHub tells a similar story from the builder’s side. Its “machine-learning” topic alone tags 219,362 repositories, and a repo can carry several AI topics at once, so the counts overlap and are never summed into one total. That is the honest reason nobody can hand you one clean “total AI projects” number: the same repo, the same underlying model, and the same wrapped consumer app can each get counted in a different tier on this page.
The pace counter adds roughly 44 tools a day as of July 17, 2026, though it is a live daily counter that swings day to day, not a stable long-run average. Watch the trend, not any single day’s counter, and always check what a source is actually counting before you quote its total.
Directory Vs Directory
Curated directories disagree because each has a different bar for inclusion. The largest counts every reviewed app; smaller editorial directories keep a tighter, hand-picked catalog.
| Directory | Listed tools | What it counts |
|---|---|---|
| There’s An AI For That | 51,242 | Every individually reviewed consumer AI tool and app. |
| Futurepedia | 4,000+ | Editorial, hand-curated catalog of notable tools. |
| OpenTools | 2,500+ | AI tools plus MCP servers, publisher-stated size. |
Frequently Asked Questions
How many AI tools are there in 2026?
There is no single number, because it depends on what you count. As of July 17, 2026, There’s An AI For That lists 51,242 consumer AI tools. But the largest machine-countable pool is 2,918,668 models hosted on Hugging Face. Each source measures something different.
Why is there no single number for how many AI tools exist?
Because “AI tool” is not one thing. A consumer app (like ChatGPT), a downloadable model, an open-source repo, and a Python package are all “AI tools” by some definition, and each source counts only its own slice. The counts also overlap, so they can never be summed into one total.
What counts as an AI tool?
This page tracks six distinct definitions: consumer AI tool, AI model, open-source AI project, AI Python package, AI dataset, and AI demo / Space. Each has its own source and its own count, laid out in the definitions table above.
How many AI models are on Hugging Face?
As of July 17, 2026, Hugging Face hosts 2,918,668 models, 962,657 datasets, and 1,422,199 Spaces (hosted demo apps).
How many open-source AI projects are on GitHub?
GitHub’s machine-learning topic alone tags 219,362 repositories (as of July 17, 2026). GitHub reports five AI-related topics separately; a repo can carry several at once, so the counts overlap and are never summed.
How many new AI tools are launched per day?
Roughly 44 per day, based on the There’s An AI For That new-tools-added-today counter on July 17, 2026. It is a live daily counter that swings day to day, not a stable long-run average.
How many consumer AI SaaS tools are there?
There’s An AI For That, the largest curated consumer directory, lists 51,242 as of July 17, 2026. Smaller editorial directories list fewer because they curate more tightly.
What is the difference between an AI model and an AI tool?
A model is a set of trained weights you download and run yourself (Hugging Face hosts 2.9M of them). A tool is a packaged app or service you use directly, usually built on top of one of those models. That is why the model count dwarfs the consumer-tool count.
How many AI Python packages are there?
PyPI lists over 10,000 packages classified under its Artificial Intelligence topic, out of roughly 660,000 total PyPI projects (as of July 17, 2026).
Which AI tool directory is the biggest?
There’s An AI For That, with 51,242 listed tools. Directory sizes differ mostly because of how tightly each one curates.
How often is this page’s data updated?
This page’s dataset was last compiled July 17, 2026. GitHub and OpenTools counts refresh each run; the other sources are verified by hand against the primary source listed in the Sources section.
Sources
Every number on this page carries a public source and an as-of date. Nothing here is an unsourced estimate.
| Source | Measures | As of | Access |
|---|---|---|---|
| There’s An AI For That | Curated consumer AI tools | Jul 17, 2026 | Public counter |
| Hugging Face | Hosted models, datasets, Spaces | Jul 17, 2026 | Listing headers |
| GitHub topics | Repositories per AI topic | Jul 17, 2026 | Free Search API |
| PyPI | AI-classified Python packages | Jul 17, 2026 | Public index search |
| Futurepedia | Editorial AI-tools directory | Jul 17, 2026 | Publisher-stated |
| OpenTools | AI tools and MCP directory | Jul 17, 2026 | Site meta |
Related Posts
AI adoption statistics for 2026: 78% of companies now use AI, the market hits $37.89B, plus country, industry, and ROI d...
What It Does ZPlatform’s AI Prompt Manager is a Chrome extension that helps you save, organize, and reuse AI prompts for...
AI product discovery jumped 4,700% in a year. This AI shopping assistant guide ranks the best tools, exact prompts, and ...
I evaluated 30+ AI SEO agencies in the US, UK, and worldwide. Here’s who’s actually delivering on GEO, LLM visibility, a...
The 125 best digital marketing agencies in India, ranked by verified Google reviews. Compare ratings, services, cities, ...
37 best places to launch your product, SaaS, or AI tool in 2026, ranked by Ahrefs Domain Rating with honest pros, cons, ...