Gamer - Real Time - Published Average: leaderboard
Table5 main values are real-time scores, not parenthetical RT-minus-time-paused deltas. Each benchmark retains its native scoring protocol; no conversion into success percentages. Trained Gamer CPT rows withheld pending exact released artifact binding. DeepSeek Vision is the explicitly cited DeepSeek-V4-Flash-Vision-Exp artifact. Metric: Native game score (higher better). Source: https://arxiv.org/html/2611.18397v1
Metric: Native game score (higher better). Source: arxiv.org. Saturation forecast: Around November 2027. 9 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol | 17.6 | #33 |
| 2 | Claude Opus 5 | 17.3 | #23 |
| 3 | GPT-5.6 Luna | 15.8 | #130 |
| 4 | GLM-5.3 Flash | 12.8 | #136 |
| 5 | DeepSeek-V4-Flash-Vision-Exp | 12.7 | #129 |
| 6 | Qwen 3.5 27B | 9.9 | #374 |
| 7 | Qwen 3.5 9B | 9.6 | #822 |
| 8 | Qwen 3.5 2B | 4.9 | #5284 |
| 9 | Qwen 3.5 0.8B | 4.5 | #8136 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=gamer-real-time-published-average · How It Works · Data refreshed daily, snapshot 2026-11-17.