全部 AI 动态

b11899

mtmd：添加 mtmd_helper_get_capabilities（#31183）* mtmd：添加 mtmd_helper_get_capabilities * 支持 audio_stream * 为保持兼容性恢复旧版 API * audio_gen_spk_ref、text_gen **网站：** - **证明文件：** - **macOS/iOS：** - [macOS Apple Silicon（arm64）](https://github.com/ggml-org/llama.cpp/releases/download/b11899/llama-b11899-bin-macos-arm64.tar.gz) - macOS Apple Silicon（arm64，已启用 KleidiAI）[已禁用](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel（x64）](https://github.com/ggml-org/llama.cpp/releases/download/b11899/llama-b11899-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11899/llama-b11899-xcframework.zip) **Linux：** - [Ubuntu x64（CPU）](https://github.com/ggml-org/llama.cpp/releases/download/b11899/llama-b11899-)

京东广告联合华为鲲鹏降低生成式广告推理时延15%，618大促完成核心场景验证

京东广告与华为鲲鹏合作，通过灵衢互联构建300TB共享内存池、重构RPC通信，实现端到端推理时延降低15%，KV Cache与RPC性能显著提升，并经2026年618大促核心场景5%流量验证；合作正向搜索、推荐等搜推场景外溢，并布局SuperCache、Agent沙箱等Agentic AI前沿方向。

推荐理由：该报道源自钛媒体（tmtpost.com）于2026-10-26发布的原创技术新闻，内容聚焦真实落地的技术协同：明确披露合作主体（京东广告/华为鲲鹏）、技术路径（灵衢互联、两级缓存、UB协议直连URMA）、量化成效（时延降15%、TP99优化超85%）、验证场景（618大促5%流量）及演进规划（SuperCache、Agent沙箱），所有关键事实均在材料中可交叉印证，无外部信息补充或主观推断。

b11898

ggml-cpu : fix unused variables in arm arch (#31180) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11898/llama-b11898-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11898/llama-b11898-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11898/llama-b11898-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11898/llama-b11898-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11898/llama-

b11897

tests : refactor test-llama-archs to not link internal libllama symbols (#30880) test-llama-archs included src/llama-arch.h and src/llama-model-saver.h, which prevented building on Windows with shared libraries since the used symbols are not exported from the DLL. - declare llama_arch_* wrappers around the internal llm_ arch query functions in the llama-ext.h staging header, together with llama_arch_kv_name() and llama_arch_supports_save() - slim llama_model_saver down to the private implementation of llama_model_save_to_file(): the fixture-writer half used by the test moves into tests/test-llama-archs.cpp as a local kv_writer - replace the roundtrip saver usage with the public llama_model_save_to_file() writing to a temp path (also removes the tmpfile() fd leak and its Windows workaround)

b11896

ci : fix race in webgpu memset (#31162) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11896/llama-b11896-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11896/llama-b11896-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11896/llama-b11896-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11896/llama-b11896-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11896/llama-b11896-bin-ub

b11895

metal : stop overflow from happening in large matrix output offsets (#30072) * Metal : stop overflow from happening in large matrix output offsets Assisted-by: Codex * metal : widen batch offsets and MUL_MAT_ID token indices Assisted-by: Codex --------- Co-authored-by: feni6 <270037094+feni6@users.noreply.github.com> **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11895/llama-b11895-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11895/llama-b11895-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/dow

b11890

deepseekv4.1: follow-up review fixes (#31137) * kv-cache : use ggml_backend_tensor_copy to refill the DSV4 K mirrors * llama : move the Engram pad id to llama_hparams and the token map to llama_model The saver no longer needs the deepseek41 model class. * deepseek41 : drop the F32 src1 request on the HC mixes matmul * ggml : add softplus TODOs in the backends that still use log(1 + exp(x)) * convert : make the DSV4/DSV4.1 modify_tensors generators, drop the redundant vision filter * deepseek41 : clean up comments * deepseek41 : drop the F32 src1 requests on the router and indexer matmuls * convert : use format_tensor_name directly in DeepseekV4Model * s/comp_kv_src/comp_il_src/g * cont : fix TODOs for log1p --------- Co-authored-by: Georgi Gerganov **Website:** - **Attestations:** - **macO

清华与港中文联合开源AKV：面向LLM智能体的KV缓存工程服务系统

清华大学与香港中文大学研究人员于2026年10月26日开源AKV——一个基于SGLang构建、专为大语言模型（LLM）智能体设计的KV缓存工程服务器。该系统通过让智能体主动丢弃并重定位KV缓存（而非传统文本历史摘要），在4个智能体基准测试中平均降低推理成本27.7%，任务平均得分提升1.7分。代码采用Apache-2.0许可证，当前支持Qwen、GPT-OSS-120B和MiniMax-M2.7等模型。

推荐理由：该成果代表LLM智能体高效推理的新范式：区别于依赖文本截断或摘要的‘上下文工程’，AKV直接操作KV缓存状态，避免重复prefill，显著缓解长程交互中的KV压力；实测显示高并发下P95延迟从229.9秒降至30.6秒，具备明确工程价值与可复现性；信息源自MarkTechPost发布的权威技术报道，发布时间（2026-10-26T07:56:45Z）与当前时间一致，属最新公开进展。

b11889

cuda : use std::string instead of VLA for archName (#30463) * cuda : use std::string instead of VLA for archName Modernize the code (old C style) and fix the following warnings: ``` ggml/src/ggml-cuda/ggml-cuda.cu:180:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension] 180 | char archName[archLen + 1]; | ^~~~~~~~~~~ ggml/src/ggml-cuda/ggml-cuda.cu:180:19: note: read of non-const variable 'archLen' is not allowed in a constant expression ggml/src/ggml-cuda/ggml-cuda.cu:179:9: note: declared here 179 | int archLen = strlen(devName); | ^ ``` Signed-off-by: Adrien Gallouët * Add a comment about archName Signed-off-by: Adrien Gallouët --------- Signed-off-by: Adrien Gallouët Signed-off-by: Adrien Gallouët **Website:** - **Attestations:** - **macOS/iOS:** - [ma

b11888

common, grammar: add recursive depth limit, make it a requirement (#31153) * common, grammar: add recursive depth limit, make it a requirement * catch more recursive functions * common : tidy up peg parser changes * grammar : reject candidates in a loop, without a depth limit The depth limit in llama_grammar_reject_candidates_for_stack counted the code points of a vocab token, so any token of 256 chars or more was always rejected under a grammar, such as the 512-space token of GLM-4.7-Flash or the long space tokens of Laguna-S-2.1. The rejection now walks the code points with its own stack of levels instead of recursing: there is no limit and no stack risk, whatever the length of a token from the GGUF, and the masks match master. * grammar : track rejected candidates as indices into the in

b11887

tests : adjust default threads for test-llama-context (#31157) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11887/llama-b11887-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11887/llama-b11887-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11887/llama-b11887-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11887/llama-b11887-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11

b11886

qwen: merge tensors on convert (#31028) * qwen: merge tensors on convert * convert: use generator functions * qwen4exp: remove conts and pre-scale **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11886/llama-b11886-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11886/llama-b11886-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11886/llama-b11886-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11886/llama-b11886-bin-ubuntu-x64.tar.gz

b11884

llama: fix exhausting CPU and MEM caused by dry sequence breaker (#31035) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11884/llama-b11884-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11884/llama-b11884-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11884/llama-b11884-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11884/llama-b11884-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/d

b11883

llama-quant : fix 4D tensor quantization using ggml_nrows() and name-based expert detection (#30820) * llama-quant : fix quantizing 4D tensors nrows_total was ne[1]*ne[2], so for a 4D tensor only the first ne[2] slices were quantized and the rest of the rows were dropped. Use ggml_nrows(tensor) to count all rows, including ne[3]. Fixes #29916 * llama-quant : detect expert tensors by name and fix imatrix size for 4D tensors The MXFP4_MOE path used ne[2] > 1 to detect MoE expert tensors. This is wrong for non-expert tensors with 3 or more dims, e.g. MLA projections (BailingMoE3) and 4D conv kernels. Use the tensor name instead: all tensors used by GGML_OP_MUL_MAT_ID are named *_exps or *_chexps. This also removes the BailingMoE3 special case. The imatrix holds one vector of ne[0] values per

b11882

grammar: limit recursion depth to prevent stack overflow (#29495) (#29755) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11882/llama-b11882-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11882/llama-b11882-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11882/llama-b11882-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11882/llama-b11882-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/

b11881

llama : respect no_alloc in all memory caches (#31087) * llama : respect no_alloc in memory caches Assisted-by: dsh:llama.cpp/ggml-org/DeepSeek-V4-Flash-Vision-Exp-GGUF (Mac Studio, M2 Ultra, 192GB) * llama : TODO: move no_alloc out of llama_hparams Assisted-by: dsh:llama.cpp/ggml-org/DeepSeek-V4-Flash-Vision-Exp-GGUF (Mac Studio, M2 Ultra, 192GB) * llama : TODO: move no_alloc out of llama_hparams Assisted-by: dsh:llama.cpp/ggml-org/DeepSeek-V4-Flash-Vision-Exp-GGUF (Mac Studio, M2 Ultra, 192GB) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11881/llama-b11881-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS

b11880

tools : reject truncated GGUF tensor data in merge (#31044) Co-authored-by: Shreya Agrawal **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11880/llama-b11880-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11880/llama-b11880-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11880/llama-b11880-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11880/llama-b11880-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/lla

b11879

CUDA: fuse FFN gate/up matmuls and GLU in MMQ with interleaved gate/up rows (#29948) * cuda: fuse FFN gate/up and GLU in MMQ with interleaved gate/up rows For prompt processing, run the gate and up matmuls of a dense FFN as one mul_mat_q launch and apply the GLU in the write-back. - The loaders take the gate weights as a second pointer and fill the weight tile with up and gate rows interleaved in chunks of 8, so each thread holds an up value and the matching gate value (rows r and r + 8 of an MMA tile). - The MMA loop is unchanged: same tile, warp layout, vec_dot and one sum[] per thread. The fused write-back computes the GLU from registers and writes half the rows. - The fused path uses the regular tiling (no stream-k) with the column tiles innermost so that neighbouring blocks share weig

b11878

ggml-cuda: fuse qwen MRoPE + set-rows (#31025) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11878/llama-b11878-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11878/llama-b11878-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11878/llama-b11878-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11878/llama-b11878-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11878/llama-b11878

b11877

vulkan: Add a larger tile size for conv3d (coopmat2 only) (#29920) * vulkan: Add a larger tile size for conv3d (coopmat2 only) This adds a larger tile size, in practice it is only for coopmat2 where we don't need shared memory to store the output (would use more than 64KB). Add some additional representative backend tests. At least, this is what codex came up with from looking at sd.cpp Wan VAE. * only apply large tile size for newer driver versions **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11877/llama-b11877-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama