llama.cpp Adds Tencent Hunyuan 3, Minimax2 Eagle3 Speculative Decoding, and SYCL Fused MoE

Tools official 1 src. ~1 min

A cluster of builds released July 13–14 expands llama.cpp's model coverage and backend performance. Build b9993 adds Tencent Hunyuan 3, a MoE architecture with per-head Q/K RMSNorm; b9990 adds Minimax2 Eagle3 speculative decoding support; b9985 ships fused top-k MoE kernels for the SYCL backend (relevant for Intel GPU users); b9986 fixes a reasoning leak in chat templates using force-opened bare templates; and b9994 (July 14) adds Metal Q2_0 quantization support.

Why it matters

Hunyuan 3 joining llama.cpp means Tencent's latest MoE model is now runnable locally, expanding the local inference ecosystem beyond the DeepSeek/Qwen family.

Importance: 2/5

Cluster of builds adding new model architecture support (Hunyuan 3), speculative decoding, and backend kernel improvements

llama-cpp inference local-ai open-source moe speculative-decoding release

Sources

official ggml-org/llama.cpp GitHub Releases — builds b9985–b9994