ModelRefs / Data Analysis Assistant — Canonical Workflow
Data Analysis Assistant — Canonical Workflow
Canonical Data Analysis Assistant workflow: text-to-SQL, code execution, reasoning models and benchmarks.
Overview
A data analysis assistant translates natural-language questions into SQL or code, runs it in a sandboxed execution environment, and explains the result in plain language, pairing a code-capable reasoning model with constrained execution and observability, a pattern closely related to coding-copilot and research-assistant workflows and often used by non-technical analysts.
Use this page to check which reasoning models and serverless-api or managed-container execution architectures fit your data environment, and which coding benchmarks such as HumanEval are relevant evidence for query-generation reliability on schemas like yours, including nested joins and time-series aggregations.
This is provisional decision support, not a guarantee of query correctness. Sandbox execution reduces but does not eliminate risk from generated code — validate outputs against known-correct queries, watch for subtly wrong aggregations or joins, and restrict write permissions in the sandbox to reduce blast radius from an incorrect or malformed query on shared production data stores and warehouses.
Implementation profile
| Category | reasoning-models |
|---|---|
| Implementation maturity | production |
| Evidence status | incomplete |
| Primary use cases | coding-copilot, extraction, reasoning |
| Deployment options | managed-api, self-hosted |
| Architectures | serverless-api, managed-container |
Candidate models with published references
- BGE-M3
- GPT-5
- GPT-5 Mini
- Claude Opus 4
- Llama 4 Scout
- DeepSeek R1
- Mistral Large 2
- Command R+
- o3
- o4 Mini
- Text Embedding 3 Large
- Claude Sonnet 4
Coverage means the model is a candidate worth evaluating for this workflow, not a ranking or a recommendation. Models whose reference pages are still in review are omitted.
Benchmarks relevant to this workflow
miracl, mkqa, mldr, swe-bench, aider-polyglot, gpqa, aime-2025, tau-bench, browsecomp-long-context, longfact-concepts, terminal-bench, mmmu, mmlu-pro, livecodebench.
Relevance is a coverage signal from the canonical registry. Each benchmark only describes its own protocol and date, so confirm the harness matches your workload before treating a score as evidence.
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Data Analysis Assistant — Canonical Workflow.