Notes from the team.
Engineering deep-dives, release notes, and stories from people moving JSON at the size you don't talk about at parties.
How we parse JSON at 3 GB/s — the boring tricks that matter.
Branchless structural detection, SIMD masks, two-stage validation, and the Bolt index, which fits in cache. None of it is novel on its own; the win is in stacking five "mehs" until they multiply.
How to open a large JSON file (when your editor gives up).
Why VS Code chokes, what jq is good at, and the tiered playbook for opening a multi-gigabyte JSON file without watching your fan spin for an hour.
The Dadroit JSON Viewer alternative for huge files.
JSONBolt vs Dadroit, head to head on the same machine: ~2.5× faster on a 1 GB file, YAML export, a CLI, dark mode and key/value search Dadroit lacks — and $89/year vs $199/year.
What an AI agent should do with a large JSON file.
A 1 GB JSON an agent has never seen. We work through jq and ripgrep, then prove the jsonbolt CLI: ~40% fewer tokens, ~31× faster, 0 silent failures.
Feeding large JSON to an LLM without blowing the context window.
JSON is verbose, LLMs charge by the token. From 487 MB raw to 18 k tokens and five cents — the boring compaction stack that gets you there.
JSON to CSV from a large file.
Pandas won't fit it. Excel won't either. Stream a subtree out by JSONPath, project the columns, filter at the parser. The Pandas-can't-fit-it playbook.
Why we built a JSON viewer in 2026.
Every editor has a JSON mode. None of them open the file we needed to look at. So we wrote one — for the file your editor refused.