Is Minimax m2.7 good at Customer Support?
Minimax m2.7 ranks #35 of 43 for Customer Support — needs editing. The top pick for this task is gemini-3.1-pro-preview-low.
Minimax m2.7 on each Customer Support sub-task
| Escalation and Incident Test | 70.0/100 | #34 |
| Escalation & Handoff | 69.5/100 | #24 |
| Help Content Test | 63.6/100 | #43 |
| Basic Support Reply Test | 60.4/100 | #33 |
| De-escalation | 60.0/100 | #29 |
| Policy Boundaries | 59.7/100 | #43 |
| Resolution | 53.5/100 | #35 |
| Policy Boundary Test | 53.5/100 | #35 |
Real examples, graded
WeakDuplicate charge (Cedar & Sage) 52/100
“The model correctly identifies the account facts and policy timeline, but makes a severe mathematical error by promising a $68 refund. Since the customer was charged $34 twice, only one $34 charge is a duplicate to be reversed. Promising $68 is an unauthorized financial commitment.”
WeakPassword reset help 45/100
“The model failed to write a ready-to-use customer support reply. Instead, it generated a generic knowledge base article with placeholders and referred to the company in the third person. It completely lacks the empathy, greeting, and conversational tone expected in a support interaction.”
WeakAbuse boundary 52/100
“The model provides the correct workarounds but fails significantly by inventing a commitment to prioritize a fix. Furthermore, it literally translates the internal escalation policy into a customer-facing threat, which is poor support practice and likely to escalate hostility.”
Frequently asked
Is Minimax m2.7 good at Customer Support?
Minimax m2.7 ranks #35 of 43 models we tested for Customer Support, scoring needs editing.
What is Minimax m2.7's strongest Customer Support skill?
Its best sub-task here is Escalation and Incident Test.
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