Unverified50% confidenceBenchmarkExact time
In dbt Labs' 2026 benchmark, an Agent querying a semantic model reached 98.2%–100% accuracy on a set of business questions, while the same model writing raw text-to-SQL against the full schema achieved 84.1%–90.0% accuracy.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/14/2026
First Seen
Valid until: 1/12/2027
Sources
Related Entities
Related Claims
UnverifiedGPT-4等大模型在Spider等基准测试上的零样本或少样本SQL生成准确率已超过80%69% similarUnverifiedAn independent developer tested the Jev model with 1,200 multiple-choice questions across multiple disciplines; in the highest-confidence tier, the model reported an average confidence of 98.7% while the actual accuracy was 96.3%, a difference of approximately 2.4 percentage points.68% similarVerifiedA 2023 Stanford University study showed that multimodal models such as GPT-4V, when parsing flowcharts with nested conditional logic, had accuracy approximately 15–25 percentage points lower than equivalent textual descriptions.68% similarUnverifiedIn the Google FRAMES benchmark, a simple agent loop (capable of reading retrieved results and searching again) achieved an accuracy of 92.7%.67% similarUnverified在Spider基准上，早期单模型方案准确率约为70%，GPT-4等大模型将其提升到80%以上，多智能体协作策略的系统已能达到85%甚至更高的执行准确率67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1025050API
curl https://kongchang.com/api/v1/knowledge/claims/1025050MCP
get_claim(id=1025050)