Four disciplines and one overall champion. Pick your lane, build for 72 hours in any language, alone or in a team of up to four, and compete for gold, silver and bronze.
Free to enter $2,000 in prizes · teams form on Discord
Kickoff in
00d00h00m00s
Sprint. How much working, useful software fits inside the 300-line budget.
Precision. Correctness under hostile input, and how well your tests cover the edge cases.
Endurance. Scope delivered end to end, and how well the parts hold together.
Freestyle. Originality of the idea and how far you took it.
Code Olympics Winter Championship18 to 21 December 2026Sprint · 300 lines, standard library onlyPrecision · ship your own testsEndurance · no size limitFreestyle · open topic$2,000 prize poolGold, silver and bronze in every laneFree to enter
Code Olympics Winter Championship18 to 21 December 2026Sprint · 300 lines, standard library onlyPrecision · ship your own testsEndurance · no size limitFreestyle · open topic$2,000 prize poolGold, silver and bronze in every laneFree to enter
Code Olympics Winter Championship. 18 to 21 December 2026. Sprint · 300 lines, standard library only. Precision · ship your own tests. Endurance · no size limit. Freestyle · open topic. $2,000 prize pool. Gold, silver and bronze in every lane. Free to enter.
3rdedition of Code Olympics
89projects judged in the first two editions
4disciplines, one rubric
72hto build
$2,000prize pool
No draw this time. You choose how you compete, and one rubric decides who wins.
Any language
Two editions assigned constraints at random. This one trusts you with the choice and judges what you do with it.
Your own lane
Sprint, Precision, Endurance or Freestyle. Each rewards a different kind of engineering, and each has its own podium.
One champion
Every project is scored on the same rubric, so the highest score across all four lanes takes the overall title.
Seg 01 · Warm-up
Why lanes, not a draw.
Code Olympics ran twice on a random draw. A draw is a fair test of range: nobody chooses, so nobody gets an easy ride. It is also a lottery, because some combinations suit a team far better than others.
Lanes keep the pressure and remove the lottery. Each discipline has one hard rule, so every team still works inside a limit, but it is a limit they picked because it fits how they build.
One rubric keeps the lanes honest. All four are scored on the same criteria, so the overall title goes to the best engineering, whichever lane it came from.
Our judges have worked at
Judges of previous Code Olympics editions.
Google
Apple
Microsoft
Amazon
Walmart
PayPal
Datadog
Snap
Bosch
New York Life
Optum
Lowe's
HCLTech
Stellar Cyber
Clinical Ink
American Family Insurance
Speekl
Quaestor Technologies
Angi
T-bank
BMA
Monaco Research
CodeMega Solutions
02 · Heats
Seg 02 · Heats
Four disciplines. Pick one.
Each lane has one rule and one thing it rewards. Gold, silver and bronze are awarded in every lane.
Lane 1 · Small and complete
Sprint
gold$300
300 lines, standard library only
Build the smallest complete tool that solves a real problem, using only your language's standard library. Blank lines and comments do not count toward the 300, everything else does. Judges reward what you fit in, not how close you get to the limit.
judged on
How much working, useful software fits inside the 300-line budget.
an idea that fits
A log summariser that reads any web server access log and prints the slowest endpoints and the error spikes, using nothing but the standard library.
what wins
Everything in the budget does real work
Handles bad input without crashing
Reads cleanly at the size it is
what loses
Minified or packed lines to dodge the count
Code generated outside the window
A demo that does one trick
Lane 2 · Correct before clever
Precision
gold$300
Ship your own test suite
Build something where being right matters: a parser, a calculator, a converter, a scheduler. Your repository includes the tests that prove it. Judges run them and then try to break your program.
judged on
Correctness under hostile input, and how well your tests cover the edge cases.
an idea that fits
A date-range parser that copes with time zones, leap years and malformed input, with a test for every case it claims to handle.
what wins
Tests run with one command
Edge cases named and covered
Wrong input fails loudly and clearly
what loses
Tests that only check the happy path
Tests written to match whatever the code prints
Untested claims in the README
Lane 3 · The full distance
Endurance
gold$300
No size limit
Build the most complete system you can in 72 hours. Several components that work together, documented and runnable from a clean checkout.
judged on
Scope delivered end to end, and how well the parts hold together.
an idea that fits
A self-hosted pastebin with accounts, syntax highlighting, expiry and an API, running from a clean checkout with one command.
what wins
Parts that work together end to end
Setup a judge can follow in minutes
Scope that is finished, not sketched
what loses
Ten half-built features
A setup that needs the author present
Screens with nothing behind them
Lane 4 · Your own brief
Freestyle
gold$300
Open topic
Bring an idea nobody asked for and make it real. Any domain, any form, as long as it runs.
judged on
Originality of the idea and how far you took it.
an idea that fits
A terminal piano that teaches chords as you play, or any idea nobody asked for, carried all the way through.
what wins
An idea the judges have not seen
Carried far enough to be used
A README that shows why it exists
what loses
Novelty with nothing underneath
A pitch instead of a program
A clone of a well-known tool
Seg 02 · Lane draw
Which lane are you?
Three questions. No sign-up, nothing stored.
03 · Final
Seg 03 · Scoring
One rubric for all four.
Scores compare fairly across lanes because the criteria are the same. Only Discipline Mastery reads differently in each.
Several judges per project
Every submission is scored independently by several judges, senior engineers from industry.
Adjusted for jury size
Scores are adjusted for how many judges reviewed each project, so a smaller field neither helps nor hurts.
Overall champion
The single highest score across all four disciplines. A lane gold and the overall title can go to the same team.
Scoring weights
30%30%20%20%
30FunctionalityIt works. Core features behave as described, the program handles bad input without crashing, and a judge can run it from the README.
30Discipline MasteryMeasured against your discipline: the line budget in Sprint, correctness and test coverage in Precision, delivered scope in Endurance, originality in Freestyle.
20Code QualityReadable structure, sensible names, clear separation of concerns. Comments that explain why, not what.
20InnovationAn approach the judges did not expect, or a familiar problem solved in a better way.
Seg 03 · Rulebook
The rules.
Short on purpose. The full brief on 14 December adds the submission checklist and how line counts and tests are checked.
Competition rules
R1
The window
All code in the submission is written between Fri 18 Dec 18:00 UTC and Mon 21 Dec 18:00 UTC. Plan, sketch and pick your stack before; the commit history should show the build.
R2
Teams
1 to 4 people. Each team enters one discipline with one project.
R3
Language
Any language. Frameworks and libraries are allowed in Precision, Endurance and Freestyle. Sprint allows the standard library only.
R4
Sprint line count
Every non-blank, non-comment line in files you wrote counts, 300 at most. Generated, lock and vendored files are excluded and must not do the work.
R5
Precision tests
A test suite in the repository that runs with one command. Judges run it, read it, then feed the program inputs of their own.
R6
Repository
A public GitHub repository with a README naming the discipline, how to run the project and the decisions behind it, plus a live link, a build or a short video.
R7
AI tools
Allowed. Someone on the team must be able to explain every part of the code, and the README says how the tools were used.
R8
Deadline
Submissions after Mon 21 Dec 18:00 UTC are not accepted. How to submit is in the brief on 14 Dec.
R9
Ownership
You keep all rights to your code.
Seg 03 · Schedule
From Discord to the podium.
All times UTC. Judging runs over the holidays at the judges' own pace, so results land in early January.
1.0
Now
Discord open
Join the Discord to follow announcements, ask questions and find teammates.
2.0
Dec 14, 2026
Full Brief Published
Detailed rules for all four disciplines, the submission checklist and how line counts and tests are checked.
3.0
Dec 18, 2026 @ 18:00 UTC
Kickoff
The clock starts. Build in your discipline, alone or in a team of up to 4.
4.0
Dec 21, 2026 @ 18:00 UTC
Submissions Close
Push your repository and submit it before the cutoff. How to submit is in the brief. Late submissions are not accepted.
5.0
Dec 22, 2026 to Jan 6, 2027
Judging
Judges score every submission asynchronously against the shared rubric. Each project is reviewed by several judges.
6.0
Jan 8, 2027
Winners Announced
Gold, silver and bronze in each discipline and the overall champion, announced on Discord with score breakdowns. Prize money is paid within 14 days.
04 · Podium
Seg 04 · Prize purse
$2,000 in prizes.
One headline prize for the overall champion and a cash gold in every lane.
Overall champion
The single highest score across all four disciplines.
Sprint gold
Precision gold
Endurance gold
Freestyle gold
Gold in each lane: cash prize, medal and certificateSilver: medal and certificateBronze: medal and certificate
Seg 04 · Recognition
Everyone who finishes leaves with something.
Every team that submitsA certificate of participation.
Gold, silver and bronze in each laneA medal and a certificate.
Lane golds and the overall championCash prizes, paid within 14 days of the results.
Every projectScores and written comments from several judges, against the published rubric.
Seg 04 · Medal table
Hall of fame.
Two editions so far, two different formats. These are the projects that took the podium.
EditionGoldSilverBronze
Summer 2026 ↗SENTINELI'm The Warrior4.312GambitGANESHA4.241QuizlympicsHANUMAN FORCE4.177
A real-time system health sentinel in TypeScript with zero imports, zero dependencies and no build step. Monitor, health checks, automation and a self-cleaner in one 605-line file.
“SENTINEL turns the no-import constraint into the core technical idea instead of treating it as a limitation.”
Iusup Izripov, judge
The 4D draw: a constraint, a line budget, a domain and a language the team did not choose. 43 teams.
2025 · Champion
Vist
CuteBunny
A visual terminal tool.
The 3D draw: random constraints, no frameworks, no libraries. 46 projects judged.
05 · Post-race
Seg 05 · Organizer
Organised by Hackathon Raptors.
A UK community interest company, 15557917, that runs online hackathons for engineers all year round.
Join the Discord, pick one of four disciplines, and build for 72 hours from December 18, 18:00 UTC. Judges score every submission. Each discipline awards gold, silver and bronze, and the highest score across all four becomes the overall champion.
Can a project count in two disciplines?
No. Each team enters one discipline with one project. The overall title is decided across all four, so no discipline is a weaker route to it.
Do I need a team?
No. Solo entries are welcome. Teams can have up to 4 people. Find teammates on the Discord.
Who judges?
Senior engineers from industry. Every project is reviewed by several judges against the same rubric.
Who owns my code?
You do. Submissions must be in a public repository so judges can review them, and you keep all rights.
How much does it cost?
Nothing. Participation is free.
Hackathon Raptors judging
Judge Code Olympics Winter Championship 2026
Constraint programming championship in four disciplines with gold, silver and bronze in each. The judging panel for Code Olympics Winter Championship 2026, 18-21 December 2026, is being formed now and is matched to this subject. To nominate someone, or to be nominated, write to us with a short note on your background.