EMRB - L3 Communication Theory: leaderboard
EMRB level accuracy for interpreting measurements in communication-theory terms for the link, solved by writing code over raw I/Q data (EMRB, arXiv 2608.24086v1, Appendix Table 5). Percent; higher is better; 13 models at temperature 0 with provider default reasoning.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 93.2 | #33 |
| 2 | Grok 4.5 | 92.3 | #54 |
| 3 | Claude Opus 4.7 | 88.3 | #51 |
| 4 | Claude Sonnet 4.6 | 80.2 | #94 |
| 5 | GLM-5.2 | 73.7 | #80 |
| 6 | DeepSeek V4 Pro | 72.8 | #112 |
| 7 | Gemini 3.5 Flash | 72.6 | #66 |
| 8 | MiniMax-M3 | 71.2 | #151 |
| 9 | DeepSeek V4 Flash | 67.5 | #155 |
| 10 | MiMo-V2.5-Pro | 59.4 | #167 |
| 11 | Gemini 3.1 Pro (Preview) | 57.8 | #64 |
| 12 | Kimi K2.6 | 48.6 | #113 |
| 13 | Llama 3.3 70B | 31.2 | #578 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=emrb-l3-communication-theory · How It Works · Data refreshed daily, snapshot 2026-11-05.