HarnessSQL: 8B Model Outperforms 32B Baselines in Long-Horizon SQL Interaction

1 reports archived1 independent sourcesupdated 11/2/2026, 17:05:00

Synthesis & Latest Updates

1 Sources Cross-Validated

HKUST and Microsoft have open-sourced HarnessSQL, the first long-horizon interactive SQL agent framework for real-world databases. By employing trajectory SFT and execution-reward RL within a sandboxed harness, it addresses the 'Train-Deploy Mismatch' inherent in traditional Text-to-SQL models. Benchmarks show an 8B parameter model achieving 45.2% execution accuracy on Spider 2.0, significantly outperforming 32B specialized models and multi-agent systems.

LATEST/HKUST and Microsoft have open-sourced HarnessSQL, the first long-horizon interactive SQL agent framework for real-world databases. By employing trajectory SFT and execution-reward RL within a sandboxed harness, it addresses the 'Train-Deploy Mismatch' inherent in traditional Text-to-SQL models. Benchmarks show an 8B parameter model achieving 45.2% execution accuracy on Spider 2.0, significantly outperforming 32B specialized models and multi-agent systems.

Hourly heat curve

2 fully observed hours

Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first

新智元 (微信公众号)T2Nov 2, 17:05·85 pts
HarnessSQL: 8B Model Outperforms 32B Baselines in Long-Horizon SQL Interaction
- Introduced HarnessSQL, a framework that resolves the lack of exploration and error recovery in static generation models via long-horizon interaction training.
- An 8B parameter model achieved 45.2% execution accuracy on Spider 2.0-SQLite, nearly tripling the baseline and surpassing 32B specialized models.
- Key methods include trajectory Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) based on execution results within a closed sandbox.
View Related Reports & Analysis Original Source