FBC Model Routing Benchmark: leaderboard
Florida Building Code routing study: 45 real code questions asked of foundation, instruction-tuned and small local models with and without RAG grounding; citation accuracy (%) is a deterministic regex match of the cited section against the published code.
Metric: Citation Accuracy (%). Source: github.com. Saturation forecast: Around 2030. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 37.78 | #90 |
| 2 | Claude Haiku 4.5 | 31.11 | #288 |
| 3 | Claude Opus 4.8 | 25.93 | #40 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #6 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=fbc-model-routing-benchmark · How It Works · Data refreshed daily, snapshot 2026-11-06.