Is claude-sonnet-4.6-high good at Structured Output?
claude-sonnet-4.6-high ranks #32 of 43 for Structured Output — usable. The top pick for this task is qwen3.7-max-low.
claude-sonnet-4.6-high on each Structured Output sub-task
| Missing & Ambiguous Data | 90.0/100 | #24 |
| Schema Adherence | 86.4/100 | #25 |
| Transformation | 85.0/100 | #29 |
| Extraction | 71.4/100 | #32 |
| Noisy Structured Output Test | 63.4/100 | #42 |
Real examples, graded
WeakAmbiguous value handling 54/100
“The model completely failed the negative constraint to 'Return only strict JSON' by appending a markdown table and text after the JSON block. It also bloated the schema with unrequested 'note' fields.”
WeakNoisy OCR extraction 24/100
“The model completely failed the strict JSON constraint by including markdown and explanatory text, rendering the raw output unparseable. Furthermore, it fabricated clean values instead of extracting the raw noisy OCR text, which the prompt explicitly penalizes.”
WeakJSON repair 55/100
“The model successfully repaired the JSON and preserved all data accurately. However, it completely failed the negative constraint to 'Return only repaired strict JSON' by including markdown formatting and an explanatory list of fixes applied.”
Frequently asked
Is claude-sonnet-4.6-high good at Structured Output?
claude-sonnet-4.6-high ranks #32 of 43 models we tested for Structured Output, scoring usable.
What is claude-sonnet-4.6-high's strongest Structured Output skill?
Its best sub-task here is Missing & Ambiguous Data.
This page is Spring Prompt, running
We just did this for every model. Do it for your prompt.
The rankings above come from running real tasks through real models and scoring every output. Spring Prompt is that same engine — pointed at your prompt, your test cases, and your definition of good.
- Generate test cases from your prompt — no eval set required to start.
- Compare models side by side with quality, cost and latency in one matrix.
- Optimise the winner until the scores say it's ready to ship.
Prompt × model results
12 test cases · 3 evals