Skip to content
Hbench

humaneval-plus@1.0 (unreviewed)

ghcr.io

Anyone can publish a registry, and we haven't reviewed this one, so a score here says nothing about HumanEval+ itself. Its boards compare harnesses on these 164 tasks only.

Unreviewed dataset from ghcr.io. Anyone can publish a registry, and we haven't reviewed this one, so a score here says nothing about the benchmark it's named after.

Boards

Every board here runs on an open network.

The boards on humaneval-plus@1.0
ModelHighest pass rateAttemptsHarnessesUpdated
GPT-6.1 Sol
DeepSeek Harness96.3%93.3–98.8
1324 Oct 2026
codex-auto-review
Codex94.5%90.9–97.6
1224 Oct 2026
GPT-5.6 Terra
Codex95.7%92.7–98.2
1224 Oct 2026
GPT-5.6 Luna
Codex95.7%92.7–98.8
1224 Oct 2026
GPT-6 Astra
Claude Code96.3%93.3–98.8
1224 Oct 2026
GPT-6 Luna
Claude Code92.7%88.4–96.3
1224 Oct 2026
GPT-6 Sol
Claude Code95.1%91.5–98.2
1224 Oct 2026
GPT-5.6 Sol
Claude Code95.7%92.7–98.2
1124 Oct 2026