Snapshot · São Paulo Grand Prix 2026 · generated 16 November 2026

F1 AI

Formula 1 race analysis, an MCP server that gives AI assistants real timing data, and a pipeline that reads telemetry off onboard video. Open source, no API keys.

São Paulo Grand Prix 2026

Race · Interlagos, Brazil · 08 November 2026

Winner
Lando Norris
McLaren
Fastest lap
1:12.689
Kimi Antonelli
Race distance
71 laps
Interlagos
Track temperature
26 °C
Rain during the race
PosDriverTeamGrid+/-GapPtsBest lap
1Lando NorrisMcLaren1Winner251:12.983
2Max VerstappenRed Bull Racing21+19+14.602181:13.131
3Isack HadjarRed Bull Racing5+2+24.485151:12.877
4George RussellMercedes7+3+33.991121:14.040
5Kimi AntonelliMercedes2-3+41.017101:12.689
6Oscar PiastriMcLaren10+4+56.47581:13.018
7Liam LawsonRacing Bulls3-4+75.40261:13.495
8Charles LeclercFerrari14+6+78.58641:14.170
9Lewis HamiltonFerrari20+11+1 LAP21:13.724
10Nico HulkenbergAudi4-6+1 LAP11:14.000
11Arvid LindbladRacing Bulls18+7+1 LAP01:14.004
12Fernando AlonsoAston Martin8-4+1 LAP01:14.706
13Franco ColapintoAlpine13+1 LAP01:13.710
14Pierre GaslyAlpine15+1+1 LAP01:14.462
15Alexander AlbonWilliams9-6+1 LAP01:14.464
16Gabriel BortoletoAudi19+3+1 LAP01:14.940
17Oliver BearmanHaas F1 Team6-11+1 LAP01:14.036
18Sergio PerezCadillac22+4+2 LAPS01:16.123
19Carlos SainzWilliams12-7+2 LAPS01:14.613
20Rafael CamaraHaas F1 Team11-9+2 LAPS01:14.369
-Lance StrollAston Martin16DNF01:17.430
-Valtteri BottasCadillac17DNS0-

Dotted lines are the second driver of each team. Hover a line for the driver and position.

Pace and tyres

Who was actually quick, once pit laps, safety-car laps and traffic-slow laps are removed.

Tyre model and strategy

Mexico City Grand Prix 2026. The São Paulo Grand Prix was run in mixed conditions, where a dry-tyre model does not apply, so this shows the most recent dry race.

Fuel effect
-26 ms/lap
the car gets quicker as fuel burns off
Pit stop time loss
23.7 s
20 stops measured
Model fit (R²)
0.80
878 clean laps
Typical error
±0.44 s
per lap

What the winner (VER) actually ran: M 27 – S 18 – S 26 laps. A least-squares model separates driver pace, tyre wear per compound and fuel burn; strategies are then ranked by predicted race time. It assumes clear air, no safety cars and linear tyre wear, with stints capped at the longest run seen on each compound.

Telemetry: the top two in qualifying

São Paulo Grand Prix 2026 qualifying. Each driver's fastest lap of the session, lined up by GPS position.

NOR · lap 23
1:27.470
top speed 299 km/h
RUS · lap 18
1:26.745
top speed 295 km/h
Gap
0.725 s
RUS faster
Full throttle
25% vs 29%
braking 21% vs 22%
CornerDistanceMin speed NORMin speed RUSTime gained by NOR
T136281.185.0-0.045
T246285.986.6-0.068
T3646172.7177.9-0.075
T41396111.2118.1-0.102
T51564153.9158.7-0.163
T62006159.1170.0-0.064
T72142157.5168.2-0.069
T8230472.565.40.068
T9243383.483.2-0.038
T10272466.168.0-0.11
T112941173.4164.20.11
T12323092.179.00.086
T133348113.5100.30.128
T143602246.7240.4-0.03
T154009285.0289.0-0.022

Sector-level gaps match official timing to a few hundredths of a second. Corner-level numbers are indicative: the public feed samples the car about four times a second.

Video to telemetry

Reading speed, gear, throttle, brake and RPM straight off the pixels of an onboard video.

Rendered onboard clip with a speed and gear dial
Test footage: NOR's lap rendered from its real telemetry, so the right answer is known.
The HUD with the regions the reader looks at outlined
Where the reader looks. A small neural network reads the digits; gauges are measured by how far their colour extends.
Frames with speed read
100.0%
750 frames
Speed error
0.01 km/h
95% within 0.1 km/h
Gear correct
100.0%
of samples
Throttle error
1.6 pts
out of 100
Brake agreement
100.0%
on / off

Measured on footage this project renders itself. Real broadcast video has a different HUD design and needs its own layout profile, calibrated in the dashboard; that has not been tested yet.

Championship

2026 standings after round 20.

The project

This page is a snapshot. The full project runs on your own machine, for any session, with no API keys.

Dashboard

A Streamlit app with six views over any session since 2023: race, pace and tyres, telemetry, video lab, championship. Everything on this page comes from it.

MCP server

13 tools that let an AI assistant such as Claude answer F1 questions with real numbers instead of guessing.

Data analysis

Clean-lap pace, a fuel-corrected tyre model, measured pit loss, strategy ranking and GPS-aligned lap comparison.

Video to telemetry

Turns an onboard video's HUD into a telemetry CSV, with a renderer that makes test footage to measure accuracy against.

Run it

git clone https://github.com/muzzkh/f1-ai.git
cd f1-ai
python -m venv .venv
.venv/Scripts/python -m pip install -e .        # macOS / Linux: .venv/bin/python
.venv/Scripts/python -m streamlit run dashboard/app.py

MCP tools

ToolWhat it does
get_scheduleRace calendar for a season: round numbers, event names, dates, which sessions each weekend has, and whether it has already been run.
get_standingsCurrent (or final) driver and constructor championship standings for a season.
get_session_resultsClassification of a session. Races: position, grid, gap to winner, status, points. Qualifying: Q1/Q2/Q3 times. Also weather and each driver's best lap.
get_lap_timesEvery lap of one driver: lap time, sectors, tyre compound and age, position, pit in/out flags and track status (GREEN / SC / VSC / RED).
get_race_paceDrivers ranked by representative pace: median of clean laps (green flag, no pit laps, within 107% of the fastest), with gap to the quickest.
get_tyre_strategyWhat each driver ran: every stint with compound, lap range, tyre age at the start and median clean pace.
get_tyre_degradationFitted race pace model: how much slower each dry compound gets per lap of age, pace offset between compounds, the fuel-burn effect, each driver's underlying pace, plus measured pit-stop time loss. A simple linear model; treat the numbers as estimates and check r2.
simulate_race_strategyRank dry-tyre pit strategies for a race by predicted total time using the fitted degradation model and measured pit loss, and show what the winner actually did. Ignores traffic, safety cars and track evolution.
compare_driversHead-to-head telemetry comparison of two laps (each driver's fastest unless lap numbers are given): lap summaries, minimum speed and time gained at every corner, and time gained in each of 25 slices of the lap. Positive gain_a values mean driver A was faster there. Corner-level numbers are indicative (about +/-0.05 s).
get_lap_telemetryTelemetry trace of one lap (fastest if no lap number), resampled to `points` evenly spaced distances: speed, throttle, brake, gear, RPM, DRS.
list_hud_profilesHUD layout profiles available for video-to-telemetry extraction.
render_onboard_videoRender a lap's real telemetry as an onboard-style video with a HUD overlay (speed, gear, throttle, brake, RPM). Useful as test footage with known ground truth for video_to_telemetry. Returns the file path.
video_to_telemetryRead telemetry (speed, gear, throttle, brake, RPM, DRS) out of an onboard video by analysing its HUD graphics, and save it as CSV. `profile` names the HUD layout (see list_hud_profiles); footage with a different HUD needs its own calibrated profile. If a reference lap is given the result is time-aligned to the official telemetry and its accuracy reported. Check speed_read_rate_pct: a low value means the profile does not fit.