how to recode emmeans::pairs() using marginaleffects

I am a fan of the marginaleffects package and really appreciate its concise and easy-to-understand syntax. Whenever I encounter the emmeans function, I prefer to convert it into marginaleffects to study it. However, I recently came across this code, and I’m stuck—I can’t figure out how to proceed. The problem is as follows:

library(tidyverse)
library(lme4)
library(emmeans)
library(marginaleffects)

d <- mtcars %>% 
  as_tibble() %>% 
  select(cyl, vs, am, gear, carb) %>% 
  mutate(cyl = cyl > 6)

m <- lme4::glmer(
  cyl ~ vs * am + (1 | gear / carb),
  data     = d, 
  family   = binomial
)

m %>%
  emmeans::emmeans(specs = ~ vs + am, type = "response") %>%
  pairs(infer = T, adjust = "tukey") %>%
  as.data.frame()

I can recode first line

m %>%
  emmeans::emmeans(specs = ~ vs + am, type = "response") %>%

as

library(marginaleffects)
m %>% 
  marginaleffects::predictions(
    newdata = datagrid(vs = unique, am = unique),
    type    = "response",
    re.form = NA
  ) %>% 
  as_tibble() 

but I can’t figure out the second line.

m %>%
  emmeans::emmeans(specs = ~ vs + am, type = "response") %>%
  pairs(infer = T, adjust = "tukey")   #<<

Could you please give me some help? Thank you so much!