# Gemini 3.5 Flash — RedlineBench results card

This is a **results-tracking model card** for the API model **Gemini 3.5 Flash**. It
holds no weights. Its purpose is to host evaluation results for
[RedlineBench](https://huggingface.co/datasets/crosbylegal/RedlineBench) on the
Hugging Face Hub, since Gemini 3.5 Flash has no public model repository.

- **Benchmark:** [crosbylegal/RedlineBench](https://huggingface.co/datasets/crosbylegal/RedlineBench)
- **Report:** https://intelligence.crosby.ai/benchmark/
- **RedlineBench `redline_overall`:** 45.1

Results live in [`.eval_results/`](./.eval_results). Scores are attributed to the
published report (community/source — not HF-`verified`, which is inspect-ai-only).