micro1 Eval Hackathon

Source ↗Official site (closed) ↗

Registration closed · Similar hackathons still open

All open hackathons →

Not sure which to join? Get 1:1 help →

AIData Developer tools

About this hackathon

A free, global online hackathon from AI data lab micro1, hosted on HackerRank. Instead of building an agent, you pick a field you know well—software, finance, law, medicine, science, design—find something today's frontier models still get wrong, and turn it into a rigorous, reproducible eval (5–10 well-graded tasks recommended). Solo only, one entry each; coding or confident use of tools like Codex or Claude Code is needed. The challenge opens Oct 30 and runs about 10 days; winners are announced Nov 23.

Status
Deadline: 2026-10-30
Starts: 2026-10-30
Who: Open to all (Open worldwide; solo only, one submission each; you need real expertise in some field plus coding skills or strong use of AI coding tools; no confidential employer or client data)
Level: Intermediate
Prize: $10k · $10,000: 1st $5,000, 2nd $3,000, 3rd $2,000; top performers may be considered for roles at micro1
Size: —
Format: Online · International
Location: San Francisco (online; placed at organizer’s city)

Original listing (Chinese)

The full original listing below is in Chinese, as published in the Hack123 newsletter.

一句话简介：AI 数据公司 micro1 在 HackerRank 上办的全球线上黑客松，⁠‌‌‍‍‍‌‌‍‍‌‌‍‌‌‍‍‌‍‌‌‌‍‍‍⁠不做 Agent，而是用你最懂的专业领域，找出前沿大模型还做不好的事，做成一个严谨、可复现的评测集（eval）。

看点：软件、金融、法律、医疗、科学、设计等任何领域都可以；建议 5—10 道评分标准清楚的题目，前沿模型「有意义地失败」最打动评委；可以先用 OpenRouter 上的免费模型调试；提交代码压缩包 + 短视频讲解 + 简介；只能单人参赛；10月30日开赛后报名入口关闭，但比赛期间仍可通过活动页的直达链接加入

奖金：1 万美元：第一名 5,000 美元、第二名 3,000 美元、第三名 2,000 美元；表现突出者有机会获得 micro1 工作机会

截止日期：10月30日 15:00（北京时间）活动页报名截止（开赛后仍可凭直达链接加入）（线上）

活动日期：10月30日 15:15（北京时间）开赛，持续约 10 天；11月23日公布结果（线上）

Summarized from the organizer’s page and public posts. Please confirm details on the official site before registering.