As far as I know, this feature is not currently supported, can you please consider this for yatai 2.0? ~~~ a_runner = bentoml.xxx.to_runner(embedded=True) b_runner = bentoml.xxx.to_runner() svc = bentoml.Service() @svc.api(...) async predict(...): a_result = await a_runner.async_run(...) # embedded runner b_result = await b_runner.async_run(...) # remote runner ( other pods ) ~~~