GameReplica - Implementation Consistency: leaderboard
Implementation consistency in GameReplica (Qiao et al., 2026): judge agents check whether the replica source code implements each rule of the reference checklist, with a first-round assessment and an independent review. Percent of rules implemented over 125 game replication tasks, three runs each; higher is better.
Metric: Implementation consistency (%; share of reference rules the replica source code implements, agent-as-judge assessment and review; 125 tasks (25 web games x 5 difficulty levels), game reached only through screenshots and mouse, keyboard and reset actions; mean of 3 runs per task; first-round judge gemini-3.6-flash, review by gpt-5.6-sol). Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 | 67.4 | #56 |
| 2 | GPT-5.5 | 39.4 | #48 |
| 3 | Grok 4.5 | 22.6 | #78 |
| 4 | GLM-5.2 | 13.6 | #137 |
| 5 | Kimi K3 | 7.6 | #53 |
| 6 | DeepSeek V4 Pro | 3.6 | #181 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=gamereplica-implementation-consistency · How It Works · Data refreshed daily, snapshot 2026-11-12.