Releases·Tool Release
llama.cpp v0.4.1 adds Maple 20B-A1B, Tencent Hy 4, Spark2.5 support and bumps ggml to v0.24.0
The llama.cpp project shipped v0.4.1 on GitHub, adding support for Maple 20B-A1B (ternary MoE, CPU), Tencent Hy 4, and Spark2.5, plus Kimi-K3 recurrent-state rollback. The release refactors JSON schema handling and chat parsing, fixes MTP KV cache and GDN normalization bugs, and bumps ggml to v0.24.0 with expanded backend coverage. It matters to anyone running local inference on the affected model families.
15 engagement1 sourceMon, Sep 14, 2026, 06:27 PM
AI Overview
- Adds Maple 20B-A1B ternary MoE (CPU), Tencent Hy 4 preview, and Spark2.5 architecture support.
- Bumps ggml from v0.23.0 to v0.24.0, adding a precision-control API and major Vulkan, SYCL, Hexagon, and OpenCL work.
- Adds Kimi-K3 recurrent-state rollback and a --fuse-qkv conversion flag for fusing Q/K/V tensors in HF-to-GGUF conversion.
- Refactors JSON schema handling via common_schema, splits chat parsers into common/parsers, and adds JSONL logging via --log-l.
- Fixes MTP context KV cache allocation for DeepSeek2 and GLM-MoE and GDN normalization from max to rsqrt for Qwen/Kimi/GLM models.
- Server changes fold child-process handling into one monitor thread, enable --reasoning-preserve by default, and fix speculative decoding after multimodal input.
- Removes deprecated --mmap/--mlock/--direct-io args in favor of --load-mode and makes mmproj and draft devices follow global --device selection.
- UI gains lazy-mount chat rendering, embedded CMake assets, and fixes MCP image attachments in tool blocks.
Related
Releases·Tool Release
llama.cpp v0.5.0 adds HRM-Text/MiMo-V2.6 support, CUDA and Metal fusion optimizations, ggml 0.25.0
ggml-org released llama.cpp v0.5.0 on 2026-09-23, focusing on backend performance and correctness, broader model coverage, and more robust server/router operation. The release bundles ggml 0.25.0, CUDA and Metal/SYCL/Vulkan optimizations, new model support (HRM-Text/DFM Mimir 1B, MiMo-V2.6, HunyuanOCR DFlash), and API changes like LoRA loading from an open FILE handle. Practitioners running local inference gain faster attention and MoE paths plus more stable multi-model serving.
ggml-org released llama.cpp v0.5.0 on 2026-09-23, focusing on backend performance and correctness, broader model coverage, and more robust server/router operation. The release bundles ggml 0.25.0, CUDA and Metal/SYCL/Vulkan optimizations, new model support (HRM-Text/DFM Mimir 1B, MiMo-V2.6, HunyuanOCR DFlash), and API changes like LoRA loading from an open FILE handle. Practitioners running local inference gain faster attention and MoE paths plus more stable multi-model serving.
1 engagement1 sourcegithubWed, Sep 23, 2026, 08:50 PM
Releases·Tool Release
llama.cpp v0.4.0 adds Qwen3.8-Flash-Next and Nemotron-3-Puzzle-75B-A9B support, lazy tensor loading, ggml 0.23.0
llama.cpp shipped v0.4.0, adding initial support for Qwen3.8-Flash-Next and Nemotron-3-Puzzle models. The release also brings on-demand tensor reading, per-slot server context limits, video input options, and a ggml update to 0.23.0 featuring major sparse flash attention and RDMA work.
llama.cpp shipped v0.4.0, adding initial support for Qwen3.8-Flash-Next and Nemotron-3-Puzzle models. The release also brings on-demand tensor reading, per-slot server context limits, video input options, and a ggml update to 0.23.0 featuring major sparse flash attention and RDMA work.
6 engagement1 sourcegithubFri, Sep 4, 2026, 07:56 PM
Social·Community
Community discovers Spark-X2.5-4B and 1.7B small LLMs with native 1M context on Hugging Face
A Reddit user browsing Hugging Face found Spark-X2.5-4B and Spark-X2.5-1.7B from XHToken, small models with their own architecture rather than fine-tunes. The 4B reportedly benchmarks neck and neck with Qwen 3.5 9B, and both claim native 1M context. They do not yet run out of the box on llama.cpp.
A Reddit user browsing Hugging Face found Spark-X2.5-4B and Spark-X2.5-1.7B from XHToken, small models with their own architecture rather than fine-tunes. The 4B reportedly benchmarks neck and neck with Qwen 3.5 9B, and both claim native 1M context. They do not yet run out of the box on llama.cpp.
242 engagement1 sourceredditTue, Sep 1, 2026, 02:35 PM
Releases·Tool Release
llama.cpp releases v0.2.0 with nightly build b10566
llama.cpp v0.2.0 is out, featuring a new nightly build and a release script for preparation. The update syncs with ggml v0.21.0 and includes CI improvements.
llama.cpp v0.2.0 is out, featuring a new nightly build and a release script for preparation. The update syncs with ggml v0.21.0 and includes CI improvements.
142 engagement2 sourcesgithub, redditFri, Aug 21, 2026, 06:32 PM
Releases·Tool Release
pi 0.81.0 integrates llama.cpp support
pi 0.81.0 now includes native support for llama.cpp via a llama-server router, allowing users to replace the huggingface/pi-llama extension or manual model management. The update is documented at pi.dev/docs/latest/llama-cpp.
pi 0.81.0 now includes native support for llama.cpp via a llama-server router, allowing users to replace the huggingface/pi-llama extension or manual model management. The update is documented at pi.dev/docs/latest/llama-cpp.
175 engagement1 sourceredditTue, Jul 21, 2026, 03:31 PM