DATA CUT 2026-07-27 116 活跃模型 680 A 类案例

CASE EVIDENCE / A RECORD

Suiundukov Mederbek / Ala-Too Inter… 使用 Grok 3 mini Reasoning (high) 处理软件工程任务执行

Suiundukov Mederbek / Ala-Too International University 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T08:13:48Z。

原始记录:Ala-Too internship project built a Grok-powered Text-to-SQL agent for Olist e-commerce analytics

A

Chinese Brief

中文案例导读

Suiundukov Mederbek / Ala-Too International University 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T08:13:48Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。 Model Atlas 不把 benchmark、教程、发布说明或集合页包装成真实案例。

厂商xAI / Grok
模型Grok 3 mini Reasoning (high)
任务类型代码代理与软件工程
审核状态auto_approved

任务

真实任务背景

这是一个围绕软件工程任务执行的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:Provide a Streamlit chat interface where non-technical users ask natural-language questions over a normalized TiDB Serverless database containing the Olist Brazilian e-commerce dataset.

代码代理与软件工程公开代码库A 类可核验real_case
公开产物

公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:The technical report describes a working end-to-end app that inspects schema, drafts SQL, checks and executes queries, synthesizes answers, and renders automatic charts for business questions such as category revenue, o…

模型作用

Grok 3 mini Reasoning (high) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作:Grok-3-mini, accessed through the xAI API with LangChain ChatOpenAI, powers the ReAct SQL agent loop: schema inspection, SQL generation, query checking, retry/refinement, and natural-language answer synthesis.

风险边界

当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:The technical report explicitly says ChatOpenAI uses Grok-3-mini via https://api.x.ai/v1; README/app comments contain an inconsistent Groq/Llama mention, so the report is used as the binding evidence. The reasoning-high…