CiCi · Restaurants

Gluten-free restaurants, scored

2,344 real venues across 803 cities. Each one can carry a number from 0 to 100 for how safely its kitchen handles gluten. The number is built from what the kitchen actually does and what happened to diners with celiac disease, never from star ratings. None of them is scored yet, and the section below the map explains why. Search by city, state, ZIP or restaurant name.

Colour — what the listing recordsentirely gluten-freegluten-free optionslimited optionsnothing recorded

Number — what CiCi has verifiedA pin shows a 0–100 safety score once a diner with celiac disease, the operator or an accreditation body has reported. Until then it shows an em dash, and the colour is a contributor's claim rather than a checked fact.

2,344 real venues, none of them verified yet

Every place on this map is real, imported from OpenStreetMap's diet:gluten_free tag, with 96 of them recorded as entirely gluten-free kitchens, across 803 cities. That tag is a contributor's note, not an audit, so none of them carries a score yet. A score needs a diner with celiac disease, the operator's own statement, or an accreditation body. Until one of those exists, saying otherwise would be inventing reassurance. We are not going to tell you “this restaurant has a shared fryer” without having checked.

You can be the first report on any of these

A single structured report from a diner with celiac disease does more for this map than a thousand star ratings, because it answers the question star ratings cannot: did anyone actually react. That is what turns a grey pin into a scored one.

A server bringing plates to a table of friends in a busy restaurant — Most restaurants have never been audited. The questions below are what close the gap.

Why the colours don't match the star ratings

This is the part worth understanding, because it is where CiCi diverges from every other gluten-free restaurant finder.

Existing apps average star ratings across everyone who reviewed a place. That breaks in two ways at once. It mixes populations: someone eating gluten-free by preference cannot detect cross-contact, because they have no symptom to report, so their five stars silently outvote the celiac diner who spent the next three days in bed. And it measures the wrong thing: star ratings capture service, price and atmosphere. A shared fryer does not lower a star rating. It is invisible in the aggregate, and it is the most common way people get glutened.

So a restaurant here can have 500 five-star reviews and still show amber. That is the map working as intended.

How the score is built

Kitchen structure: up to 50

Dedicated fryer, separate prep area and utensils, fresh pasta water, whether wheat flour is used on the premises at all. A fully gluten-free kitchen takes the whole allocation, because it is the only structural guarantee that exists in food service.

Third-party audit: up to 20

GFCO, GFFP, Coeliac UK, AIC and equivalent schemes that actually inspect. A published allergen matrix earns part of it. A restaurant describing its own practices is not the same claim as a body with a published standard having checked them.

Staff competence: up to 12

Reported by celiac diners who asked directly. Staff who understand cross-contact score higher than staff who only know the word “gluten”.

Outcome evidence: up to 18

Did diners with celiac disease actually react. This is the only component that can pull the score down hard, because what happened to people outranks what the kitchen says it intends.

Three rules that can only lower a score

⚠️ Unknown never counts as safe
A fact nobody has confirmed scores zero, not partial credit. A venue with no celiac reports and no accreditation is capped at 45 no matter how good its published policy reads. The page says why it was capped.
⚠️ Evidence caps the number
Two reports cannot produce a 95. Accreditation relaxes the cap, because a certification is checkable without any diners at all. Popularity does not relax it, because a pile of reviews from people who cannot detect cross-contact is not evidence about cross-contact.
⛔ Reported reactions override paperwork
Where a meaningful share of celiac diners report reacting, the score is capped at 25 to 40 regardless of stated policy, training or allergen matrix. No amount of documentation outranks what happened to people who ate there.

Highest scoring right now

Check it in the aisle instead

Point CiCi at the barcode or check a restaurant before you go and get this same reasoning in about two seconds, plus the carbs and whether it has been recalled. 360,000+ packaged foods are already read and waiting.

Try it free for 14 days →

Browse by city

What to ask when the score can't tell you

Most restaurants have never been audited and never will be. Ask these three questions instead. The answers resolve most of the risk in about ninety seconds.

Do you have a dedicated fryer, or is the oil shared with breaded food?The most common single point of failure. It also covers fries, chips and anything battered in one question.
Can this be prepared on a clean surface with clean utensils?Covers boards, tongs, grills and the pass in a single ask.
Is wheat flour used in the kitchen?Airborne flour settles across a kitchen and stays on surfaces for hours. A pizzeria answers this differently from a steakhouse.

And say “I have celiac disease”rather than “I eat gluten free.” Kitchens respond to those two sentences very differently, and the word “cross-contact” signals that you mean the process rather than the recipe.

The full guide to eating out gluten freeWhat actually goes wrong in restaurant kitchens, and the timing and phrasing that reduces it.Gluten-free travelCountry rules, which places are easy, and why a good translation card explains cross-contact rather than naming gluten.

The scanner

Reading a page is fine. Standing in the aisle is the hard part.

This site answers the question you knew to ask. The app answers the one you didn't: the barley malt in a rice cereal, the wheat starch in a gummy. Scan the barcode, or photograph the panel when the barcode isn't in any database, and CiCi shows which words on the label produced the answer.

360,000+ packaged foods already read. Ingredient statement, allergen declaration and nutrition panel, each one assessed by the same rules you can see working on this page, so most scans are an instant answer rather than a lookup.
Unlimited scanning, about two seconds an answer. Fig and Spoonful both cut you off at five scans a month.
Carbs and the full nutrition panel on every scan. Total and net carbohydrate with the arithmetic shown. For the roughly 6% of people with type 1 diabetes who have coeliac disease too, that is two apps collapsed into one.
Your glutening log. Record a reaction and CiCi shows what those days had in common. It counts overlaps and never tells you what caused it.
Saved products. The regular shop in one place, so next week is not a re-scan of the same forty items.
Scanning happens on your device. No scan is ever attached to you, sold, or sent to an analytics service.

Try it free for 14 days →See pricing

14 days free, then $29.99 a year. Cheaper than Fig, and it does what three separate apps do. This website stays free forever either way: every answer on it, permanently, with no account.