pracadata/mexico-fuel-prices

Mexico Fuel Prices by Municipality (daily, CNE)

Retail gasoline (regular and premium) and diesel prices in Mexico, by municipality, product and day, from the price file that every fuel permit holder reports to the national energy commission (CNE, formerly CRE). One row per municipality × product × day with the number of stations and, when there are at least 3 stations, the median, minimum, maximum and quartiles of the pump price in MXN and USD. A second table says, for every day, how complete the capture was and how many rows were rejected. Daily since 2026-10-25: 5 days so far, 29 October 2026 is the latest. The archive grows every day; the official CNE feed only shows the current day and keeps no history, so a day that was not captured cannot be recovered. Updated weekly: the latest day in this build is 2026-10-29.

Palabras clave: precios de gasolina por municipio, precios de la gasolina y el diésel en México, gasolina regular y premium, CNE, Comisión Nacional de Energía, precios por municipio, diésel.

Per-station prices in MXN, USD and EUR: on the Apify Store (pracadata/mexico-fuel-prices). Pay per row, with an API and MCP for AI agents. See also the Brazilian equivalent, pracadata/brazil-fuel-prices (weekly, ANP).

Read this first: coverage and limitations

• Short history. The series starts on 2026-10-25, the first day this project captured the CNE feed, and has 5 days in this build. There is no earlier history and none can be reconstructed. Do not read trends into a few days of data.
• Prices are reported by the permit holders and are not audited. The CNE publishes what each station reports. Typing errors, stale prices and filler values exist; this dataset only removes the obvious ones (see the next point). Do not use it as an official price reference.
• Rejected rows, counted every day (fuel_coverage_daily_mx). Per day, in this build: 19 price rows outside 5 to 60 MXN per litre (so far all of them filler values such as 0.01 or 1), 18 to 20 rows from stations that report two different prices for the same product on the same day (n_rejected_duplicate_divergent: the station and product are dropped for that day, because there is no way to know which price is right), and 111 exact duplicate rows that are collapsed into one (n_collapsed_duplicates: nothing is lost, the price stays). The accepted ones are about 37,276 price rows per day from 13,848 stations.
• About 0.6% of the stations in the CNE registry could not be placed in a municipality (their coordinates are 0, 0, empty, or far from any municipality: wrong at the source). Almost all of them report no price at all, so among the stations with a price, 4 (0.029%) are missing from the municipality table on the latest day. The municipality is derived by us from the station's coordinates (point in polygon on the INEGI municipal boundaries), not reported by the CNE: a station with wrong coordinates can land in the wrong municipality.
• Ciudad de México appears as its 16 alcaldías (the INEGI boundaries split it), each one a municipality here.
• Sale price to the public only (price at the pump, per litre). No purchase price, no taxes breakdown, no other fuels (no LPG, no CNG).
• At least 3 stations to publish a statistic. With fewer, only n_stations is filled and the price columns are null.
• Mexican pesos, nominal. price_usd_* converts the MXN price with the ECB euro reference rate of the previous business day (USD/MXN is a cross rate), indicative only; the rate itself is never published.

The day of a price: reference_date

The CNE feed is updated once a day, at 18:00 Mexico City time (UTC-6, Mexico has no daylight saving time since 2022), and shows only the current day. A price file captured between 18:00 on day D and 17:59 on day D+1 belongs to day D: reference_date = date(capture time in UTC-6 − 18 hours). If the capture ran several times in that window, the last version wins. A capture at 07:05 on 26 October is therefore the price of 25 October.

coverage_flag describes the day: complete, partial (fewer stations than 80% of the median of the previous 7 days with data) or gap (a calendar day with no new capture of the file). In this build: 5 complete, 0 partial and 0 gap days. The first days have no history to compare with and are complete.

What is in this repository

File (config)	Rows	Content
data/fuel_price_municipality_daily_mx.parquet (fuel_price_municipality_daily_mx)	22,355	municipality (code, name and state) × product × day, 17 columns, 1,547 municipalities
data/fuel_coverage_daily_mx.parquet (fuel_coverage_daily_mx)	5	one row per day with stations, prices, rejects by reason and coverage_flag, 12 columns
DATA_DICTIONARY.md		every column: type, definition, where it is null; products and units
manifesto.json		files, row counts, SHA-256 of each file, date of the data build, sources

Products: gasoline (regular), gasoline_premium and diesel, all in MXN per litre.

Examples

Median gasoline price by state on the latest day, as the median of the municipality medians (each municipality counts once, whatever its number of stations), in DuckDB:

select state, round(median(price_mxn_median), 2) as gasoline_mxn_per_litre, count(*) as municipalities, sum(n_stations) as stations
from 'hf://datasets/pracadata/mexico-fuel-prices/data/fuel_price_municipality_daily_mx.parquet'
where reference_date = date '2026-10-29' and product = 'gasoline' and price_mxn_median is not null
group by 1 order by 2 desc;

The same in pandas (needs pyarrow and huggingface_hub installed):

import pandas as pd
df = pd.read_parquet("hf://datasets/pracadata/mexico-fuel-prices/data/fuel_price_municipality_daily_mx.parquet")
last = df[(df["reference_date"] == df["reference_date"].max()) & (df["product"] == "gasoline") & df["price_mxn_median"].notna()]
print(last.groupby("state")["price_mxn_median"].median().sort_values(ascending=False).round(2))

from datasets import load_dataset
ds = load_dataset("pracadata/mexico-fuel-prices", "fuel_coverage_daily_mx")

Source and attribution

• Fuente: Comisión Nacional de Energía (CNE), precios reportados por los permisionarios (Acuerdo A/041/2018). The price file the permit holders must report under agreement A/041/2018 of the Comisión Reguladora de Energía, published by the CNE: https://www.cne.gob.mx/ConsultaPrecios/GasolinasyDiesel/GasolinasyDiesel.html.
• Fuente: INEGI, Marco Geoestadístico, diciembre de 2025. Municipal boundaries, names and codes, under the INEGI "Términos de Libre Uso de la Información" (commercial use allowed with credit to INEGI as author): https://www.inegi.org.mx/inegi/terminos.html. The assignment of each station to a municipality is our own derived work, not an INEGI product, and INEGI has not endorsed it.
• Fuente: Banco Central Europeo, euro reference exchange rates (used only for the USD columns): https://www.ecb.europa.eu/stats/policy_and_exchange_rates/euro_reference_exchange_rates/html/index.en.html.
• Licence and terms of use. The licence field is other because this is official public data of the Mexican government, redistributed with attribution. The CNE page publishes no licence or terms of use for the price file. This dataset is unofficial and not affiliated with the CNE or INEGI. Check the source terms for your use.
• Takedown / corrections: write to lucashpissaia@gmail.com. If the CNE or INEGI ask for it, the data is removed. This repository has no personal information: prices are aggregated by municipality; station names, permit holders' names and addresses are not published here.

How to cite

@dataset{mexico_fuel_prices,
  title  = {Mexico Fuel Prices by Municipality (daily, CNE)},
  author = {pracadata},
  year   = {2026},
  url    = {https://huggingface.co/datasets/pracadata/mexico-fuel-prices},
  note   = {Derived from CNE and INEGI public data. Fuente: CNE, precios reportados por los permisionarios (Acuerdo A/041/2018); Fuente: INEGI, Marco Geoestadístico, diciembre de 2025}
}

Terms

• Provided as is, without warranty, from public sources, with the limitations above. Not official; do not use as a regulated price reference without checking the CNE.
• Cite the sources above. If a publisher restricts reuse, the data may be removed from this repository going forward.
• Questions or corrections: open a discussion on this repository. Takedown requests: lucashpissaia@gmail.com.