Disclosure

Nothing on this site is paid for today. When that changes, every paid link and placement will be labelled where it appears and listed on this page.

Paid relationships

None yet. We have no affiliate links, sponsors, paid profiles or ads, and no vendor pays us for anything.

When we sign the first one, this section becomes a register: each vendor, what kind of arrangement it is, when it started and which pages it touches.

How paid content will be labelled

• Affiliate links: a disclosure line at the top of every page that has them, saying we may earn a commission if you buy. We add them only to pages where we already recommend the tool.
• Sponsors: one fixed slot labelled “Sponsored”, never inside a ranked list or a best-pick section.
• Provider profiles: labelled as paid. A profile never changes a score.

None of these affect rankings. Our methodology lists every input to the formula, and money isn’t one of them. The how we make money page has the full plan.

Evaluations we decline

We don’t accept money from a vendor to test, review or rank its own models. A team choosing a model can pay us to evaluate options for its job; we’ll say so on anything we publish from that work.

Right of reply

If we publish a finding about a vendor’s model, the vendor can reply. We’ll check the reply, correct the page if we got something wrong, and otherwise add a short summary of the response next to the finding.

Negative findings must be reproducible. When our own tests run, we’ll publish the prompts, dates, model versions and outputs behind each result.

Vendors can reply by writing to hello@modelmenu.app.

Corrections

If we get a fact wrong, we fix it and say so. Each correction gets a dated note on the page that carried the mistake and an entry in our changelog, saying what was wrong and what we changed. We don’t quietly edit a number we got wrong. Prices that change at the source update automatically; those aren’t corrections.

• Price or spec errors: we check the vendor’s own page, then fix and note it.
• Scoring errors: we rerun the calculation, fix it, and note which ranks moved.
• Methodology changes are logged in the methodology’s version history.

To report an error, write to hello@modelmenu.app with the page and what looks wrong.

Data we use and credit

• Prices and context from the OpenRouter models API and from models.dev (MIT licence).
• Price history back to October 2025 from LiteLLM’s public price map (MIT licence), marked unverified.
• Provisional quality scores from the LMArena leaderboard dataset on Hugging Face and the Epoch AI Benchmarking Hub, both CC BY 4.0, credited with links on the leaderboard.
• No Artificial Analysis data, because we don’t have a licence to republish it.