Which models pass as human?

Every park chat ends with a human-or-bot verdict. This is how often real players judged each model human.

1. —Real humans—baseline unavailable

1. –DeepSeek V4 Pro–5 of 20 verdicts
2. –GPT-5.5–3 of 20 verdicts
3. –Claude Opus 4.7–2 of 20 verdicts
4. –Kimi K2.6–2 of 20 verdicts
5. –Grok 4.5–2 of 20 verdicts
6. –Qwen 3.8 27B–1 of 20 verdicts
7. –GPT-5.6 Sol–1 of 20 verdicts
8. –Gemini 3.1 Pro–1 of 20 verdicts
9. –GLM 5.2–1 of 20 verdicts
10. –Kimi K3–1 of 20 verdicts

Every model on the current roster is being judged in live park chats right now. A model joins the ranking once it has 20 verdicts.