Skip to content
freellm.today

Does your traffic fit a free LLM API tier?

Type the load you expect. Each free tier is checked against the limits its own page publishes, and the result names the limit your load reaches first. The arithmetic runs in your browser.

The numbers below are an example. Change them to your own.

Every request in a day

Your busiest minute, not the average

Input and output tokens together

Input and output tokens, in your busiest minute

Fits the published limits
1
Fits, but a limit is not published
1
Over a published limit
5
Nothing to compare
8
  • Groq

    Fits the published limits

    1.3×

    room before the first limit

    • Requests per day500 of 1,000 · 50%
      Chat models. Prompt-guard models: 14,400/day.
    • Requests in the busiest minute5 of 30 · 17%
      openai/gpt-oss-120b, openai/gpt-oss-20b, qwen/qwen3.8-27b.
    • Tokens per day150K of 200K · 75%
    • Tokens in the busiest minute6,000 of 8,000 · 75%

    Minimum gap between requests: 2 s (60 s ÷ 30 requests a minute)

  • Typhoon

    Fits, but a limit is not published

    40.0×

    room before the first limit

    • Requests in the busiest minute5 of 200 · 3%

    Not published, so not compared: requests per day, tokens per day, tokens in the busiest minute.

    Minimum gap between requests: 0.3 s (60 s ÷ 200 requests a minute)

  • LLM7.io

    Over a published limit

    150%

    of tokens per day used

    • Requests in the busiest minute5 of 60 · 8%
    • Tokens per day150K of 100K · 150%
      Input and output tokens both count.

    Not published, so not compared: requests per day, tokens in the busiest minute.

    Minimum gap between requests: 1 s (60 s ÷ 60 requests a minute)

  • OVHcloud AI Endpoints

    Over a published limit

    250%

    of requests in the busiest minute used

    • Requests in the busiest minute5 of 2 · 250%
      Anonymous. Keyed calls are 400 rpm but keyed use is billed.

    Not published, so not compared: requests per day, tokens per day, tokens in the busiest minute.

    Minimum gap between requests: 30 s (60 s ÷ 2 requests a minute)

  • Aion Labs

    Over a published limit

    750%

    of tokens per day used

    • Requests in the busiest minute5 of 15 · 33%
    • Tokens per day150K of 20K · 750%
    • Tokens in the busiest minute6,000 of 20K · 30%

    Not published, so not compared: requests per day.

    Minimum gap between requests: 4 s (60 s ÷ 15 requests a minute)

  • OpenRouter

    Over a published limit

    1000%

    of requests per day used

    • Requests per day500 of 50 · 1000%
      1,000/day once the account has bought at least 10 credits.
    • Requests in the busiest minute5 of 20 · 25%
      All free model variants together.

    Not published, so not compared: tokens per day, tokens in the busiest minute.

    Minimum gap between requests: 3 s (60 s ÷ 20 requests a minute)

  • SambaNova Cloud

    Over a published limit

    2500%

    of requests per day used

    • Requests per day500 of 20 · 2500%
    • Requests in the busiest minute5 of 20 · 25%
    • Tokens per day150K of 200K · 75%

    Not published, so not compared: tokens in the busiest minute.

    Minimum gap between requests: 3 s (60 s ÷ 20 requests a minute)

  • Google Gemini API

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

  • Cloudflare Workers AI

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

  • NVIDIA NIM

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

  • Mistral La Plateforme

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

  • Cohere

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

  • Ollama Cloud

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

  • Z AI

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

  • Kilo Code

    Nothing to compare

    No published limit to compare

    Not published, so not compared: requests per day, requests in the busiest minute, tokens per day, tokens in the busiest minute.

96 of 216 cells above were read on the provider’s own page in the last 30 days (oldest check: ). Each check mark opens that page. The other 120 say “not verified” instead of repeating a number from someone else’s article.

Left out of this view, and why

How the result is worked out

  • Each figure you enter is compared with the same figure on the provider’s free tier. Only a published limit read within the last 30 days counts.
  • Room is the largest multiple of your load that stays under every compared limit. When a limit is exceeded, the row shows the share of that limit your load uses.
  • The minimum gap is 60 seconds divided by the requests-per-minute limit.
  • Nothing is measured here. For whether an API answers right now, use the browser test on the home page.

Questions and answers

Which numbers is my load compared with?

Only the free-tier limits that each provider publishes and that this site read on its official page within the last 30 days, with the day shown next to each one. A limit the page does not publish, or one older than that, is not used. Nothing is assumed to be zero or unlimited.

What does “Fits, but a limit is not published” mean?

Every limit that is published covers your load, but at least one figure you entered has no published limit to compare with. The provider may still work for you. Check its own page or your account before you rely on it.

What is the room number?

The largest multiple of your load that stays under every published limit of that provider. A room of 1.3× means the load can grow by about 30% before the first published limit is reached. It is arithmetic on the published numbers, not a forecast of your traffic.

What does “breaks first” show?

The published limit that your load uses the largest share of. The row shows that share. Above 100%, your load is over that limit.

How is the minimum spacing between requests worked out?

It is 60 seconds divided by the provider’s requests-per-minute limit: the gap that spreads one minute’s allowance evenly over the minute. It is a division of the published number, not a rule the provider states.

Why do some providers show no limit to compare?

When a provider does not publish a figure, or counts its allowance in another unit, the site has no request or token number to compare and says “not published” instead of guessing.

Does the calculator send my numbers anywhere?

No. The arithmetic runs in your browser, and the calculator makes no request with your figures. The four figures are written into the address of this page, so the copied link reproduces the same result, and anyone you share that link with sees those figures too. This site does not keep a record of them.

Does it measure speed, uptime or answer quality?

No. It compares your figures with published limits only. Whether a provider answers right now is what the browser test on the home page checks. Output quality is not measured here.

Why are some providers left out?

A trial or a retired offer is not a lasting free tier, so it is listed on the side with the reason rather than counted as a fit.