Question
Given the completed quality and resource/reliability evidence, which exact runtime, model revision, quantization, prompt format, context limit, default/optional role, and M4 Pro/24 GB recommendation should enter Tab’s initial Local Model Catalog and implementation-ready migration spec?
Select only combinations that pass every predeclared gate. Apply the material-improvement rules when comparing passing combinations, preserve inconclusive results as unsupported rather than inferred support, and do not implement production local inference in this ticket.
Question
Given the completed quality and resource/reliability evidence, which exact runtime, model revision, quantization, prompt format, context limit, default/optional role, and M4 Pro/24 GB recommendation should enter Tab’s initial Local Model Catalog and implementation-ready migration spec?
Select only combinations that pass every predeclared gate. Apply the material-improvement rules when comparing passing combinations, preserve inconclusive results as unsupported rather than inferred support, and do not implement production local inference in this ticket.