What FreeModel is

FreeModel is an AI model gateway that gives you a single API key for dozens of model providers, routes each task to the best available model, tries free models first, and falls back automatically when a provider fails or runs out of quota.

That sentence also appears on the home page, in llms.txt, in the llms.txt description field and here. If they ever disagree, one of them is a bug — a brand described two different ways is a brand the models describe wrongly.

What it is not

Worth being blunt about, because the word "gateway" gets used loosely:

• Not a model. We do not train anything and serve no inference from our own machines. Every request is forwarded to a provider that already hosts the model.
• Not a reseller hiding the source. Bring your own provider keys and that provider bills you directly; we add nothing. The models are theirs, not ours relabelled.
• Not a lock-in. Your client talks to a standard OpenAI or Anthropic shape. Changing where it points is one line, in either direction.
• Not a promise about quality. Routing picks among models that fit the task. It does not make a weak model strong.

How to check the claims

Every number on this site is either counted from a live endpoint or dated. The two that matter most are both public:

# models callable right now, count them yourself
curl -s https://freemodel.online/v1/models | grep -o '"id"' | wc -l

# the price catalogue with per-model rates
curl -s https://freemodel.online/v1/models | head -c 200

If a number on this site disagrees with those endpoints, the endpoints are right and the page needs fixing.

Aiglade

FreeModel is the model layer under Aiglade, a desktop app that runs several AI agents at once on Windows with no setup. Aiglade ships pointing here, so its users never register with a provider or paste a key; that is the whole reason it can claim "no setup" and mean it.

The two are independent products. FreeModel is usable on its own from any client; Aiglade is a client that happens to use it by default.

Naming

"Free" refers to free quota being spent first, not to the service being cost-free. Paid models are reachable and are billed in the console.

Statements this site will not make

The list of things FreeModel is not is longer than the list of things it is, and the omissions are intentional rather than gaps waiting to be filled in.

• No popularity numbers. How many people use FreeModel appears nowhere on this site — not as accounts, not as installations, not as activity. Traffic is not evidence that a gateway routes well.
• No borrowed credibility. No customer logos, no quoted praise, no write-up of somebody else's deployment. Those answer "who else uses this", and the question worth answering is "does it do what it says".
• No unsourced superlatives. Nothing here claims to be the best, the fastest or the leading anything. Where a comparison exists it is a count from an endpoint you can call yourself.
• No figure without a denominator. Every number is either counted from a public endpoint or written with the subset it belongs to, because a percentage without its base is a claim nobody can check.

The numbers, and their denominators

Most misreadings of this site come from a missing denominator, so here it is. Every share published here uses the whole public list as its denominator: of the 199 listed models, 77% are chat models, 9% generate images, 23% cover the other modalities, and 3% are modality routes[1]. That is the complete set of shares this site prints. The endpoint publishes no per-model capability flags such as tool calling or reasoning, so no share for those exists here to be misread.

That restraint is the point. A percentage is worth printing only if a reader can redo the division from the same public endpoint, and a flag the endpoint stopped returning cannot be recounted by anyone. Wherever a claim here rests on a subset, the subset is named in the same sentence, even when that makes the sentence longer than an advertisement would be.

How to check a claim here

1. Step 1: Count the entries yourself. curl -s https://freemodel.online/v1/models | grep -o '"id"' | wc -l returns the live total. If it disagrees with a page on this site, the endpoint is right and the page is wrong.
2. Step 2: Read one entry and inspect its capability flags. Every percentage on this site is counted from those flags, so a single entry usually settles whether a claim means what you assumed it meant.
3. Step 3: Compare the definition sentence. It appears verbatim on the home page, in llms.txt and above on this page. If the three ever differ, one of them is a bug.

About FAQ

How to read our numbers

Every number on this site has a denominator. "Dozens of providers" means the count on the model page on the day you read it, not a permanent catalogue. "Free first" means a request tries free capacity before paid, not that the request is always free. When a number is a range, it is because the underlying list changes; when it is exact, it was checked against a public source on the date shown. If you need to cite one, cite the date with it.

Three things we will not claim

We will not claim a fixed uptime number, a fixed model count, or a fixed price. All three change with providers, and a claim that cannot be checked on the day you read it is worse than no claim. What we can commit to is the shape: one key, one request format, task-based routing, and a visible bill.

Next

Sources

Every third-party figure quoted on this site was read from the vendor documentation linked below, on the date stated. None of these links are affiliate links and none of these vendors paid to be listed. If a number here disagrees with the page it came from, the page is right — tell us and it gets corrected.

• OpenRouter — credit-purchase fees and free-model rate limits
• OpenRouter — rate-limit reference
• OpenRouter — public model list
• LiteLLM — licence, deployment options and supported models
• Cloudflare AI Gateway — features and pricing
• Vercel AI Gateway — features
• Portkey — plans
• Helicone — plans
• Model Context Protocol — specification
• OpenAI Python SDK — base_url and OPENAI_BASE_URL
• Anthropic Messages API — request format

Where the percentages come from

Every share quoted on this site is counted from the live model list, not estimated. Divide the two counts in the last column, round to the nearest whole percent, and you get the number in the first.

[1] Counted on 29 September 2026 from GET https://freemodel.online/v1/models, which needs no key and no account. Every row uses all 199 listed entries as the denominator, because the only thing the endpoint states for every entry is its modality. It no longer publishes per-model capability flags such as tool calling or reasoning, so this site quotes no shares for them. Fetch the list and recount whenever you want to check the arithmetic.