Back to browse
GitHub Repository

⚡ A blisteringly fast, zero-allocation JSONL search engine in Zig. Query and aggregate massive datasets at 4.0 GB/s using SIMD-accelerated "Blind Scanning." Up to 50x faster than jq.

42 starsZig

Zog parsing JSONL at 3 GB/s by not parsing it

by aikoschurmann·Feb 22, 2026·2 points·0 comments

AI Analysis

●●●BangerWizardryBig Brain

Skips JSON parsing, hits 20–40x faster throughput than jq on real benchmarks.

Strengths
  • Zero-allocation streaming architecture with partial parsing is genuinely inventive—not just faster, it's smarter
  • Rigorous benchmarks against ripgrep and jq; aggregations (COUNT, SUM) at ~2 GB/s rival SQL tools
  • Fills a real gap: jq is slow but flexible; this trades flexibility for speed on JSONL, a common format
Weaknesses
  • Query language is terse and operator-heavy; no piping to other tools limits composability
  • Early stage: no release beyond GitHub, no package manager distribution yet
Target Audience

Data engineers, log analysts, JSONL-heavy workflows

Similar To

jq · ripgrep · DuckDB

Similar Projects