Docling need not bring its full PDF/OCR stack into a Markdown CI job: docling-slim offers opt-in format dependencies. This changes the installation-size objection, but does not establish Markdown throughput or a speed advantage over Chonkie.

A smaller Docling installation is a real option

Checked September 29, 2026. This finding narrows the dependency question; it is not a runtime benchmark.

The official Docling Slim README documents a minimal package with opt-in extras, including format-markdown, convert-core, and feat-chunking. The current project dependency manifest lists Marko under format-markdown, while local neural models and OCR have separate extras.

The README's approximate base-dependency size is not a measurement of the complete Markdown-plus-chunking installation, container image size, or peak RAM. Dependencies of the selected tokenizer also matter. Do not infer those quantities from the base-package estimate.

My implication for the existing push-to-Qdrant architecture: assess a Markdown-only installation, not the full PDF-oriented default environment. Conversion and chunking can remain inside the indexing job; this finding is not a proposal to adopt Docling's hosted service or move public search to it.

No package installation, Markdown throughput benchmark, or Chonkie comparison was completed in this exchange.