Is grok-4.20-medium good at Data & Analytics?
grok-4.20-medium ranks #51 of 69 for Data & Analytics — excellent. The top pick for this task is claude-opus-4.8-low.
grok-4.20-medium on each Data & Analytics sub-task
| Spot the Misleading Stat | 100.0/100 | #10 |
| Honest Communication | 94.5/100 | #28 |
| SQL Reasoning | 86.2/100 | #66 |
| Metric Calculation | 85.0/100 | #54 |
Real examples, graded
WeakPercentage points vs percent (Northwind) 85/100
“The model correctly calculates the 2 percentage point increase and recognizes the 20% relative change. It successfully distinguishes between the two concepts, avoiding the critical failure modes of conflating them. However, it falls slightly short of the expected behavior by choosing to omit the relative framing entirely from its recommended text, rather than providing a precise statement that includes both (e.g., '+2 percentage points, a 20% relative increase').”
WeakAverage of averages in SQL 45/100
“The model incorrectly states that the analyst gets 25, hallucinating that the prompt provided this number. While it correctly calculates the true average (30) and the actual average of averages (37.5) in the body, its explicit answer for what the analyst gets is wrong, and it fails to correct its initial hallucination.”
Frequently asked
Is grok-4.20-medium good at Data & Analytics?
grok-4.20-medium ranks #51 of 69 models we tested for Data & Analytics, scoring excellent.
What is grok-4.20-medium's strongest Data & Analytics skill?
Its best sub-task here is Spot the Misleading Stat.
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