TeleResilienceBench - srsRANBench: leaderboard

TeleResilienceBench (2026) on its srsRANBench subset (srsRAN software-stack understanding): the model sees the question, the options and the first half of a flawed reasoning trace that led a weak generator to a wrong answer, and must continue it to the correct option. Score is the percent of the 47 instances recovered (correct flip rate); higher is better.

Metric: Correct flip rate (CFR, %): share of the 47 srsRANBench instances (srsRAN software-stack understanding, GSMA Open-Telco ot-lite) where the model, given the question, the options and the first half of a flawed reasoning trace from qwen3.5:2b that ended in a wrong answer, continues the reasoning in its thinking stream and returns the correct option; one continuation prompt, Ollama builds on one RTX 4090; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 8 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1nemotron-3-nano:4B (Ollama)46.8
2gemma4:31B (Ollama)44.7
3gemma4:26B (Ollama)44.7
4gemma4:e2B (Ollama)36.2
5gemma4:e4B (Ollama)25.5
6qwen3.5:9B (Ollama)23.4
7qwen3.5:27B (Ollama)19.1
8qwen3.5:4B (Ollama)17

Interactive version: theaggregate.ai/benchmark?slug=teleresiliencebench-srsranbench · How It Works · Data refreshed daily, snapshot 2026-11-21.