b10514 b10509
ggml-org/llama.cppAug 20, 2026 → Aug 20, 20260 days between releases
Generating AI comparison
Analyzing what changed between b10514 and b10509. This page updates automatically — usually within a minute.
The Two Releases
b10514
b10514
Aug 20, 2026· 2 months ago
Introduces support for new Granite model architectures featuring Sliding Window Attention (SWA), including GraniteSWAForCausalLM and GraniteMoeSWAForCausalLM, along with infrastructure for converting rope patterns.
Full release notesb10509
b10509
Aug 20, 2026· 2 months ago
Adds a new parameter `ggml_rope_set_offset` to control Rotary Positional Embedding (RoPE) offsets, with full support implemented for CPU, Metal, CUDA, and Vulkan backends.
Full release notes