Search results

[NEW FEATURE] Add multimodal support

enhancementNew feature or requestNew feature or request

Status: Open.

#9 In stmay4/llama.csharp;

· stmay4 opened on Jul 3, 2026

[BUG] Qwen 3.5–3.6 models produce gibberish output when using Vulkan backend with model in VRAM and KV cache offloaded to RAM

bugSomething isn't workingSomething isn't working

Status: Open.

#4 In stmay4/llama.csharp;

· stmay4 opened on Jun 30, 2026