SchemaBridge Round Trip - SchemaBench Referential F1 Detailed: leaderboard

SchemaBridge evaluates relational-schema design, description and reconstruction. Experiment-specific cohorts retain their own populations. Graph edit distance, hallucination, structural F1, text similarity and identifier coverage are distinct measurements. Round-trip runs use a fixed GPT-4o Mini constructor with each named describer. This board reports micro structural Referential F1 (0-1). Higher is better.

Metric: Micro structural Referential F1 (0-1). Source: arxiv.org. Saturation forecast: Around February 2027. 5 models tracked. Updated 14 Nov 2026.

Top models

#ModelScoreOverall rank
1GPT-5.4 Mini0.909#398
2Claude Sonnet 4.60.869#165
3deepseek-coder-6.7B-instruct0.629#7426
4CodeLlama-13B-Instruct-hf0.485#9043
5Llama 3.1 8B Instruct0.428#4545

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=schemabridge-round-trip-schemabench-referential-f1-detailed · How It Works · Data refreshed daily, snapshot 2026-11-16.