v3.1.3 v0.4.0

Name: llama.cpp
Author: ggml-org

low risk

ggml-org/llama.cppSep 4, 2026 → Sep 4, 20261 day between releases

What's New

Older release notes provided are for paperless-ngx, not llama.cpp. Newer release adds Qwen3.8 support and lazy tensor reading.

New Features

Initial support for Qwen3.8-Flash-Next and Nemotron-3-Puzzle models
Lazy tensor reading via --lazy-mode to optimize memory usage
Per-slot server context limits for better resource management
Enhanced video input options and multimodal processing improvements

Improvements

Update to ggml 0.23.0 featuring sparse flash attention and Apple RDMA transport
Bumped session/state versions for KV-cell token tracking
Added n-gram history lookup

Upgrade Notes

Cannot upgrade from paperless-ngx to llama.cpp as they are different software products.

The Two Releases

v3.1.3

Paperless-ngx v3.1.3

Sep 4, 2026· 2 months ago

A maintenance release focusing on UI fixes, AI suggestion improvements, and backend stability.

Full release notes

v0.4.0

Sep 4, 2026· 2 months ago

llama.cpp 0.4.0 introduces support for Qwen3.8-Flash-Next and Nemotron-3-Puzzle models, adds lazy tensor reading and per-slot server context limits, and updates ggml to 0.23.0 with sparse flash attention and RDMA capabilities.

Full release notes