Codebuff solves at least 40% of issues on SWE-Bench by March 31, 2025
➕
Plus
1
Ṁ1250
Mar 31
16%
chance

(This market is AI-generated but I read it and it seems right)

This market predicts whether Codebuff will achieve a 40% success rate on the SWE-Bench dataset, which is a benchmark of human-selected program issues.

Resolution will be based on official results published on the SWE-Bench dataset or Codebuff project's official channels.

References:

Get
Ṁ1,000
and
S3.00
Sort by:

Nice. We can do it!

I assume you mean the full SWE bench. We're more likely to work on the Lite or Verified subset.

© Manifold Markets, Inc.Terms + Mana-only TermsPrivacyRules