Anthropic’s original performance take-home, now a community leaderboard. Hand-optimize a kernel for a simulated wide-issue VLIW machine and chase the lowest cycle count.
Your perf_takehome.py builds the instruction stream; the judge scores the worst cycle count across nine seeded validation runs — lower is better. Model reference runs on the board show where the frontier sits. Check out the GitHub repo to get started.
Community project — not affiliated with or endorsed by Anthropic.
Ranked by clock cycles (lower is better) — each account’s best score counts
| Rank | Author | Cycles | Attempts |
|---|---|---|---|
| #1 | @redswimmer_ai | 895 | 7 |
| #2 | @zartbotF | 908 | 1 |
| #3 | @isenbaev | 916 | 6 |
| #4 | @justinwetch | 920 | 13 |
| #5 | @adrianleb | 921 | 13 |
| #6 | @josusanmartin | 923 | 109 |
| #7 | @samaltman43 | 924 | 29 |
| #8 | @yevhenx | 933 | 37 |
| #9 | @dougallj | 950 | 34 |
| #10 | @cameronspirals | 956 | 8 |
Model rows are reference runs shown for orientation only — single data points under specific harness conditions. Replication attempts may yield different results.