NCP-Bench
Models narrate interactive stories while preserving facts, commitments, and plot order against player interventions. A model auditor checks 100 movie-derived environments after each turn. The evaluation reports survival, conflicts, and narrative progress.
The player, narrator, and auditor have separate roles. The reference methods share a common evaluation interface. Auditor settings differ between the main comparison and the HiAgent experiment.
- Released
- August 2026
- Kind
- Multi-turn conversation
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Subject
- Languages and literature
- Language
- English
- Interaction
- Multi-agent
- Environment
- A simulator
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Agents
- Strongest model tested
- GPT-5.2 #51 on the capability index
- Last model added
- January 2026
- Scores refreshed
- November 20th, 2026
In the authors’ words
The rapid advancement of Large Language Models (LLMs) is revolutionizing AI for Games by enabling open-ended and fluid interactive storytelling. However, existing research has largely overlooked the critical challenge of maintaining long-horizon logical consistency and narrative integrity against unconstrained user interventions. To address this, we formulate this challenge as Narrative Commitment Preservation (NCP), and take interactive narrative as our testbed. We introduce NCP-Bench, a benchmark of 100 narrative environments derived from movie synopses. Each environment includes a structured narrative specification (trajectory, commitments, and initial facts) that we can automatically check throughout the interaction between the player agent and the narrator agent. Experiments across state-of-the-art LLMs reveal a substantial long-horizon consistency gap: high linguistic quality does not guarantee commitment preservation; even strong models frequently generate logically conflicting content under adversarial interventions, with the best-performing model (GPT-5.2) achieving only 42% survival rate after 20 turns and fact conflict rates ranging from 40% to 68% across models, and only isolated runs satisfying all achievement commitments within the 100-turn limit.
GPT-5.2 leads with 32.9 among models released by January 2026 GPT-5.2 leads with 32.9 among models released by January 2026
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in January 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 GPT-5.2 Gemini-2.5-Flash auditor 32.9
- 2 GPT-4o-mini (snapshot unspecified) GPT-5.4-mini auditor 30.1
- 3
Qwen3-235B-A22B Gemini-2.5-Flash auditor 16.8
- 4
DeepSeek-V3.2 Gemini-2.5-Flash auditor 15.9
- 5 Grok 4.1 Fast Gemini-2.5-Flash auditor 7.87
- 6 Kimi K2.5 Gemini-2.5-Flash auditor 2.88
Source: Official NCP-Bench leaderboard
Source: Official NCP-Bench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.2 | 32.9 | Gemini-2.5-Flash auditor | Dec 2025 | Official leaderboard |
| 2 | GPT-4o-mini (snapshot unspecified) | 30.1 | GPT-5.4-mini auditor, HiAgent | Official leaderboard | |
| 3 | | 16.8 | Gemini-2.5-Flash auditor | Apr 2025 | Official leaderboard |
| 4 | | 15.9 | Gemini-2.5-Flash auditor | Dec 2025 | Official leaderboard |
| 5 | Grok 4.1 Fast | 7.87 | Gemini-2.5-Flash auditor | Nov 2025 | Official leaderboard |
| 6 | Kimi K2.5 | 2.88 | Gemini-2.5-Flash auditor | Jan 2026 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.2 | 32.9 | Gemini-2.5-Flash auditor | Dec 2025 | Official leaderboard |
| 2 | | 16.8 | Gemini-2.5-Flash auditor | Apr 2025 | Official leaderboard |
| 3 | | 15.9 | Gemini-2.5-Flash auditor | Dec 2025 | Official leaderboard |
| 4 | Grok 4.1 Fast | 7.87 | Gemini-2.5-Flash auditor | Nov 2025 | Official leaderboard |
| 5 | Kimi K2.5 | 2.88 | Gemini-2.5-Flash auditor | Jan 2026 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official NCP-Bench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official NCP-Bench leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Instructions and chat
- IFEval In use Models answer about 500 prompts containing verifiable instructions such as word limits or required keywords. Code checks each instruction for strict and loose accuracy.
Gemma 4 31B IT 98.9%
- IFBench In use 58 new out-of-domain verifiable output constraints. Scored by programmatic constraint checks (strict or loose). Flint 89.2%
- MultiChallenge In use Multi-turn conversations testing instruction retention, inference memory, versioned editing and self-coherence. Graded by LLM judge with rubrics. GPT-6.1 Sol 82%
- Multi-IF In use Models follow verifiable instructions over three-turn conversations in eight languages, extending IFEval. Scored by instruction accuracy at each turn. LFM3-1.7B 89.1%
- AdvancedIF In use Models respond to over 1,600 prompts with complex, multi-turn, and system-level instructions. A judge model checks each response against expert-written rubrics.
MAI-Thinking-1 85%
- MM-IFEval In use Multimodal models follow image-grounded instructions with exact output constraints. Checks mix rule-based verification and a judge model. GPT-4o 64.6%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 10 months ago, in January 2026, for Kimi K2.5. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (6 entries).
- Pass Models scored 5 models we track, plus 1 other entry on its leaderboard.
- Warning Newest score The newest came 10 months ago, in January 2026, for Kimi K2.5.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on github.com: 5 rows for models we track and 1 other entry at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.