ττ-bench Leaderboard

pronounced “hyper-tau-bench”

RankDeveloperDateHarnessReasoning
Human performance
82.2%
#1
Claude Opus 5
Aug 31, 2026🛠️ Claude Codemax
23.9%
#2
GPT-5.6-sol
Aug 31, 2026🛠️ Codexxhigh
22.0%
#3
GPT-5.6-terra
Sep 1, 2026🛠️ Codexxhigh
18.0%
#4
Kimi K3
Sep 2, 2026🛠️ OpenCodemax
17.9%
#5
Kimi K3
Aug 31, 2026🛠️ Kimi Codemax
16.1%
#6
Claude Sonnet 5
Sep 2, 2026🛠️ Claude Codemax
14.9%

Submit Your Results

Have new results to share? Submit your developer configuration — coding harness and builder model — with a pull request that adds a submission.json under hyper-submissions/. The README's submission section has the required format and process.