Help: Choosing the optimal Cas9 vector (px330 vs pSpCas9(BB)-2A-Puro) for homozygous knockouts in mouse ES cells

I've been setting up a CRISPR workflow to generate homozygous knockouts in murine embryonic stem (ES) cells, and I'm torn between two plasmid backbones that keep popping up in the literature:

• pX330 (Addgene #42230) – the classic U6‑sgRNA drive with a bare‑bones backbone and a neomycin resistance cassette.
• pSpCas9(BB)-2A‑Puro (PX459) (Addgene #62988) – same U6 promoter but adds a self‑cleaving 2A peptide linked to puromycin resistance.

Both have been reported to work, but my lab's conditions are a bit particular:

• Need strong selection to eliminate non‑transfected cells, since our ES cells are notoriously fickle.
• Prefer a vector that leaves minimal scar after knock‑in, because we plan downstream recombination with a floxed reporter.
• Concerned about off‑target activity; some papers suggest that the type of promoter driving Cas9 (EF1α vs CBh) can affect fidelity.

What have you found in practice?

• Efficiency of homozygous indel formation (any numbers are welcome)
• Toxicity or growth impacts on ES cells
• Ease of plasmid preparation (some kits recommend mini‑prep versus midi‑prep)
• Any tweaks you made – e.g., adding a GFP reporter, swapping puromycin for blasticidin, or using a modified Cas9 (high‑fidelity SpCas9‑HF1) in the same backbone.

I’m leaning towards the puromycin vector because I’ve had decent success with puromycin selection on fibroblasts, but I’m hesitant about the 2A peptide possibly affecting protein expression in ES cells.

Any detailed anecdotes, protocol tweaks, or even a simple recommendation would be greatly appreciated. Thanks for your thoughts! ??