NCP-Bench: leaderboard

Metric: Average turns until termination (out of 100): mean over the environments of the turns played before the first confirmed conflict, the 100-turn limit or full commitment satisfaction; 100 movie-synopsis narrative environments, each with an initial fact ledger, reference trajectory and commitment set; the evaluated model narrates against an adversarial Gemini-2.5-Flash player, and Gemini-2.5-Flash auditors check every narrator response and end the episode at the first confirmed fact, commitment or player-input conflict, at 100 turns, or when every achievement commitment is met; temperature 0.6, top-p 0.95. Source: arxiv.org. Saturation forecast: Around 2038. 6 models tracked.

Top models

#ModelScore
1GPT-5.232.92
2GPT-4o Mini24.8
3Qwen 3 235B A22B16.76
4DeepSeek V3.215.88
5Grok 4.1 Fast7.87
6Kimi K2.52.88

Interactive version: theaggregate.ai/benchmark?slug=ncp-bench · How It Works · Data refreshed daily, snapshot 2026-09-29.