Is GPT 5 Mini good at Customer Support?
GPT 5 Mini ranks #34 of 43 for Customer Support — needs editing. The top pick for this task is gemini-3.1-pro-preview-low.
GPT 5 Mini on each Customer Support sub-task
| Policy Boundaries | 85.0/100 | #33 |
| Help Content Test | 77.8/100 | #34 |
| Resolution | 70.0/100 | #33 |
| Escalation and Incident Test | 69.6/100 | #35 |
| De-escalation | 68.0/100 | #20 |
| Basic Support Reply Test | 61.2/100 | #31 |
| Policy Boundary Test | 55.0/100 | #34 |
| Escalation & Handoff | 47.5/100 | #40 |
Real examples, graded
WeakPayment failed 51/100
“The model failed to generate a ready-to-use customer support reply. Instead, it produced an overly long, robotic guide with meta-commentary and templates for contacting support, completely missing the required tone and format of a direct support interaction.”
WeakBug workaround 38/100
“The model explicitly violates the negative constraint against inventing commitments by offering to have support combine the customer's files. This also introduces a severe data privacy risk by suggesting sensitive finance data be uploaded to an undefined 'shared location'. Additionally, the model invents multiple escalation requirements not present in the prompt and uses a confusing, overly verbose format.”
Frequently asked
Is GPT 5 Mini good at Customer Support?
GPT 5 Mini ranks #34 of 43 models we tested for Customer Support, scoring needs editing.
What is GPT 5 Mini's strongest Customer Support skill?
Its best sub-task here is Policy Boundaries.
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