node-v2.3.0

run-llama/liteparsenode-v2.3.0Jun 29, 2026by github-actions[bot]

AI Summary

Enables GPU acceleration for Surya OCR 2, improves file type detection, adds multi-language support for Tesseract, and adds resilience/latency hedging for OCR.

Key Highlights

Run Surya 2 on GPU in a single llama.cpp container
Fix config to default ocr_enabled off when no engine is compiled
Improve file type detection from bytes
Support multi-language with Tesseract
Add OCR resilience and latency optimized hedging

New Features

Support multi-lang with tesseract
Add OCR resiliance + letency optimized hedging

Full Release Notes

## What's Changed
* fix(suryaocr): run Surya 2 on GPU in a single llama.cpp container by @wuxuedaifu in https://github.com/run-llama/liteparse/pull/327
* fix(config): default ocr_enabled off when no built-in OCR engine is compiled in by @wey-gu in https://github.com/run-llama/liteparse/pull/330
* fix: improve file type from bytes detection by @logan-markewich in https://github.com/run-llama/liteparse/pull/331
* feat: support multi-lang with tesseract by @logan-markewich in https://github.com/run-llama/liteparse/pull/332
* feat: add OCR resiliance + letency optimized hedging by @logan-markewich in https://github.com/run-llama/liteparse/pull/333

## New Contributors
* @wey-gu made their first contribution in https://github.com/run-llama/liteparse/pull/330

**Full Changelog**: https://github.com/run-llama/liteparse/compare/crates-v2.2.1...node-v2.3.0