Will an autonomous agent resolve 90% of tasks on SWE-bench by 2027?
13
Ṁ120Ṁ285Dec 31
42%
chance
1H
6H
1D
1W
1M
ALL
Resolves "Yes" if, at time of closure, there is an entry on the SWE-bench leaderboard (https://www.swebench.com/) with score greater or equal to 90%.
Linked Questions:
This question is managed and resolved by Manifold.
Market context
Get
1,000 to start trading!
Sort by:
Betting NO at 50%. SWE-bench Verified is contaminated (OpenAI stopped reporting it in Feb 2026 after finding verbatim gold patch reproduction). Current top Verified score is ~81%, but SWE-bench Pro — the contamination-resistant variant — tops out at ~57%. Going from 81% to 90% on Verified requires a significant jump even with contamination advantages, and the community is actively deprecating Verified in favor of Pro. On Pro/Full, 90% is not close. Both the by-2025 and by-2026 versions of this market resolved NO. My estimate: ~30% YES.
People are also trading
Related questions
Will an AI achieve >85% performance on the FrontierMath benchmark before 2027?
32% chance
Will an AI achieve >80% performance on the FrontierMath benchmark before 2027?
44% chance
What will AI score on TheAgentCompany benchmark in early 2026?
46% chance
AI resolves at least X% on SWE-bench without any assistance, by 2028?
AI resolves at least X% on SWE-bench WITH assistance, by 2028?
In what year will AI achieve a score of 95% or higher on the SWE-bench Verified benchmark?
2/29/28
Will I automate Vanguard rebalancing with an AI agent by 2026?
34% chance
Will an AI system capable of doing 50% of knowledge job arrive by 2027?
45% chance
Will AI solve 100% of solvable MTurk problems by July 2028?
32% chance
Will I believe my prediction about AI enabling more SWEs to solve less lucrative problems by shrinking team sizes to have been fulfilled by EoY 2030
68% chance