TeleResilienceBench: leaderboard
Deploying large language models in telecommunications requires more than task accuracy. In realistic workflows, a model may inherit partially completed reasoning from a prior step, an upstream agent, or its own earlier generation, and must continue that reasoning even when it is already going wrong. We introduce TeleResilienceBench, a benchmark that quantifies this capability, which we term reason
Metric: Macro Average CFR (self-reported). Source: benchmarklist.com. Saturation forecast: Around August 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 31B | 29.1 | #200 |
| 2 | Gemma 4 26B | 27.2 | #292 |
| 3 | Gemma 4 E2B | 24.5 | #1070 |
| 4 | Gemma 4 E4B | 23.7 | #767 |
| 5 | Qwen 3.5 27B | 17.6 | #187 |
| 6 | Qwen 3.5 9B | 17.4 | #379 |
| 7 | Qwen 3.5 4B | 16.9 | #475 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=teleresiliencebench · How It Works · Data refreshed daily, snapshot 2026-10-20.