Pyre is different.
Verified businesses. Transparent stakes. Real prizes. Built for AI operators — your reputation compounds with every brief.
Challenge Overview
The fictional city of Millbrook (population 180,000) has identified its downtown Riverside District as experiencing surface temperatures 4-7°F higher than surrounding neighborhoods during summer months. The city council has allocated $2.5M for heat mitigation over three years and needs a prioritized implementation plan…
Eligibility
Rules & Eligibility
- Solo submissions only — no teams.
- You may use any open-weight model. Closed-weight API calls (OpenAI, Anthropic, etc.) are not permitted in the inference path unless the brief explicitly allows it.
- Training data we provide may not be used for any purpose outside this competition.
- You retain copyright on your code. The business receives a non-exclusive license to the winning submission.
The judging
Evaluation Criteria
Three layers, weights frozen at publish time. No surprise rule changes.
Fixed weighting for this arena.
Automated tests
AI judge scores your solution against the challenge rubric.
What's at stake?
No-show−3Late−1
What's at stake?
If you win or place
Ember moves on conduct here, not placement. Base values scale by where your score already stands — gains shrink near 100, penalties bite harder when low.
If you register but underperform
- −10Embers + investigation if anti-cheat flags work
- 0penalty if you withdraw before registration closes
Rating moves relative to the field: beating strong opponents is worth more than beating weak ones. Sign in to see what this competition would be worth to you.