v1.0.0

Blog post: https://huggingface.co/blog/ggml-org/llama-cpp-v1

Overview

llama.cpp v1.0.0 removes the deprecated public API and promotes the NextN/speculative API to llama.h, adds OCP FP8 KV caches, a host-memory MoE expert cache with a configurable CPU share, and loading of external PLE shards. It adds three text architectures (Kimi-K3.1-Pro, Ripple and xing4_0), drama-3 TTS, and MageVL-Cowork and Kimi-K3.1 vision, hardens llama-server (SSRF, CORS, DNS rebinding, MCP tool cancellation) and adds a /v1/audio/speech endpoint, ships a redesigned web UI with chat groups and a markdown editor, and updates ggml to v1.0.0.

API changes

• llama.h: remove llama_set_warmup, 35 deprecated functions, defrag_thold and --defrag-thold (#30565)
• llama.h: promote the NextN API from llama-ext.h and rename llama_set_embd_state_src to llama_set_ctx_state_src (#30565)
• llama.h: add moe_cache_size and moe_cache_cpu_share to llama_context_params (#30966)
• mtmd.h: add MTMD_GEN_AUDIO_TYPE_DRAMA3 and temp in mtmd_helper_gen_audio_inp (#31613, #31655)
• mtmd-helper.h: add mtmd_helper_resampler for streaming raw PCM at the model sample rate (#31964)
• server: add the OpenAI-compatible /v1/audio/speech endpoint and a temperature option for /tts (#31763, #31655)

New models

• Kimi-K3.1-Pro: K3 with the full-attention MLA layers replaced by DSA (#31779)
• Ripple-65B-Preview (ricursive): new dense 65B architecture, no new ops (#31692)
• xing4_0: mHC, MLA, MoE and MTP (#31589)
• drama-3 TTS (fishaudio) (#31613)

Core changes

• FP8 KV caches (-ctk/-ctv f8_e4m3) with optional static per-tensor scales (#28392)
• MoE cache: CPU share for misses, LRFU/ClassLRU replacement, no CPU wait on cached experts (#30966, #31233, #31888)
• external PLE/engram loading from separate GGUF shards via --shard/--ple (#31185)
• Windows unbuffered model load, mirroring the Linux DirectIO path (#26014)
• reuse the backend scheduler and graph results across reserves, and skip unused graph metadata pages (#32009, #32089, #32043)
• mark every sampling chain as a graph output to prevent output aliasing (#32015)
• remove the non-fused GDN paths and the ubatch-dependent graph node budget (#32078)
• store the MTP conv snapshots of every recurrent model in one set_rows (#31660)
• per-model build_rope with a KV-shifting fix and tests (#31683); the model is a member of llm_graph_context (#31733)
• fix on-device state copy sizes for quantized KV caches (#30541)
• follow the reference precision of ggml_prec_set_acc call sites (#31851)
• attach NVFP4 FFN scales in Granite and qwen2 (#31449, #31657); merge tensors on convert for qwen3.5 (#32102)
• apply the iGPU lazy-mode fallback only when mmap is off (#31957)
• fit: account for skipped NextN layers when setting n_gpu_layers (#31569)
• jinja: chained comparisons and argument unpacking (#32072); TTS text normalization via the chat template (#32053)
• chat: honor response_format in the qwen3-coder/deepseek/ministral3, Muse Spark and gemma4 parsers (#31956, #32017, #31915)
• json-schema: fix allOf of plain types (#31834); escape control characters and sanitize invalid UTF-8 in tool arguments (#31937, #31861)

Multi-modality changes

• remove clip.h and consolidate the model/subcontext infrastructure into mtmd (#31865)
• share the compute buffer among subcontexts to save memory for vision+audio+gen models (#31945)
• compute log-mel in the graph instead of spawning threads per request (#31808)
• support Gemma 4 streaming audio input (#31965)
• support MageVL-Cowork (#31819) and Kimi-K3.1 vision (#31980)
• require the projector and patch embedding tensors and a 2x2 spatial merge at load (#32129, #32130)
• require an F32 position embedding for HunyuanVL and read the MiniCPM-V width from ne[1] (#32131, #32132)

Server changes

• security: add --allow-unauthenticated-tools and --media-allow-private-urls opt-ins (#32113, #32115)
• security: fix --cors-origins, DNS rebinding, CORS proxy SSRF and child instance access (#32098, #32115, #32120)
• restrict the HF_TOKEN to huggingface.co hosts (#31730)
• cancel MCP tool calls on client disconnect or timeout and notify the MCP server (#32025)
• keep the server awake while a request uses its state with --sleep-idle-seconds (#32133)
• use reference-counted context checkpoints instead of deep copies (#30137)
• visit parent slots before child slots for shared-prefix requests (#30779)
• realtime API: send error events instead of crashing, reject non-object session updates, fix the output lang (#31632, #31714, #31982)
• accept the top-level response_format.schema form and convert Responses text.format JSON schemas (#28697, #26145)

UI changes

• sidebar: fold, preview and rearrange sections (#31936) and gather chats into named groups (#32012)
• move Models, MCP and Settings to a sidebar footer, with full-screen pages on phones (#32022)
• mobile: native layout, safe areas and full-screen dialogs (#31949)
• open the virtual keyboard in the chat editor on Android (#32060)
• chat input: live-preview markdown editor that expands for long messages (#31955, #31973)
• open the actions menu on right-click for chats, tabs, messages and models (#31545)
• PDF attachments: show text and page thumbnails, stop endless conversion retries (#31468)
• Discover: show audio capability from the mmproj (#31837) and keep downloads during search (#31467)
• add a search_chats tool (#32051) and move the JS sandbox setting out of developer (#32054)
• take the logo and favicons from the llama-brand package (#32039, #32112)
• fix a hidden sidebar intercepting clicks (#31712, #32035)

ggml changes

• v1.0.0: ggml 1.0.0 prunes the deprecated public API and the legacy DIAG_MASK ops, adds configurable YaRN truncation and FP8 support, and brings broad performance work across backends: few-row quantized matmuls and MoE fusion on CUDA/Metal, a Vulkan prefill FlashAttention path, a device-side MoE scheduler on SYCL, decomposed Adreno FlashAttention in OpenCL, new WebGPU kernels, expanded Hexagon op coverage, and scheduler improvements that avoid uploading unused inputs.

Assets

Nightly build: b12359

More info

• Releases and versioning of ggml-org projects

Changelog since v0.9.0

b1b770d90 examples : remove explict version from test-cmake (#32191) 4cda6227a llama.cpp : bump version to 1.0.0 (#32190) 685981335 sync : ggml 3bfde28f9 ggml : bump version to 1.0.0 (ggml/1690) 28d37fe2e examples : clean-up old/obsolete stuff (#32184) bd2767d06 ci: fix HIP quality check by ignoring the FP8 vec FA kernel spill (#32183) 82503b42f ci : set windows-cuda ccache size to 1G (#32185) 5676dfbe0 chat : accept as K2 Horizon closer at medium/low effort (#30335) 0a66ac718 server, fix CI: read a small temp file in the read_file tool test [no release] (#32168) 892b40b91 server: allow private media URLs in the vision tests [no ci] (#32167) e97010147 server: skip LAN address tests when the host cannot reach it [no ci] (#32166) 92d6c8a5c fix(hexagon): make host build work with clang-cl on Windows ARM64 (#32011) 8ca5a8008 llama : prune (#30565) 29dbbad2b ggml : prune (#29801) 2a4daf0e0 hexagon: add Q5_0 and Q5_1 support (#32094) d8c8cc4d9 server : cancel MCP tool calls on client disconnect or timeout (#32025) 3ba1ce0e2 qwen3.5: merge tensors on convert also (#32102) 0fb8f1577 Title: CUDA: few-row MMQ for BS 2-24 (#31533) 827466b9e ci : build ui on non-slim runner (#32138) 8954f1948 server: fix SSRF for CORS proxy and media URLs (#32115) a1a68eb79 server: explicit opt-in tools via non-loopback addr, check cors host port (#32113) e182fdad4 server: disallow direct access child instance from browser (#32120) 09452cded server: keep the server awake while a request uses its state (#32133) cfb9d4b75 mtmd: require an F32 position embedding for HunyuanVL (#32131) 72f4603c7 mtmd: read the MiniCPM-V projection width from ne[1] (#32132) 8c388364e mtmd: require the MLP projector and patch embedding tensors (#32129) 08f49598c mtmd: require a 2x2 spatial merge for the Qwen-VL projectors (#32130) a8575a3ef ggml : route GATED_DELTA_NET to the SpacemiT RVV kernel (#31897) a963f6595 vulkan : Intel FA kernel for prefill. (#29357) ab4a77b9a ggml-cpu: use the 4x4 simd_gemm tile on ARM NEON (#32128) 0d69255c5 CUDA: Optimize prefil via fuse of w_s scale in epilogue MMQ for nvfp4 checkpoints (#26048) 3bce8bd50 ggml-backend : prefetch only inputs computed by earlier splits (#32100) 81f21f4b2 ggml : dispatch SpacemiT Q4_1 32x256 through the HP kernel (#31929) b19f1d9e6 ui: add search_chats tool (#32051) d90cfd4a9 ui: install llama-brand from a GitHub tarball instead of git (#32112) 60d46a5ab server: when --tools-runtime is specified, ignore x-tool-runtime (#32119) 93cf736a8 chat : refactor gemma4 to build loose schema rules and clean up (#31915) 4fffab623 llama : avoid touching unused graph metadata pages (#32043) 09338b85e sycl: make oneDNN FA cache thread-safe and memoize runtime failures (#31138) dcbff2166 sycl : fix AOT for more than 4GB (#31873) f3d7725a7 sycl : use cached device-id in set_main_device (#32013) 7021f7e09 CUDA: add cuda debug buffer pool (#31919) e0949d09b opencl: add bin kernel kernel_gemm_noshuffle_q4_0_f32_32b_trans_ila_a7_bin (#30947) 5d4c8eeb2 CUDA: skip empty MoE tiles in MMQ stream-k fixup (#32067) dd6b945be server: use the moved test model in the router CORS test (#32117) e1f7593c3 opencl: add decomposed FA for Adreno (#32046) 7bcbc04fb ui: move js sandbox setting out of "developer" (#32054) dd00306bc docs: update image link in README.md (#32038) 1d8199278 ui: keep downloads in Discover during search and fix their progress with a draft (#31467) bd717e08f ui: take the logo and favicons from the llama-brand package (#32039) 2cbe0ec42 server: fix --cors-origins handling + protect against DNS rebinding (#32098) 664f031f4 ci: move test models to ggml-org-dev (#32096) 4180722ee metal : prevent out-of-bounds reads in SOLVE_TRI kernel (#32103) 055c3c26e vulkan: support batched tensor copy API (#31418) 2656b71a1 vendor : update LibreSSL to 4.4.1 [no ci] (#32106) cb44fdb18 llama : reuse graph result instances across scheduler reserves (#32089) cd2db800d jinja : support chained comparisons and argument unpacking (#32072) f7b886fe6 chat: allow TTS model to provide custom text normalizer via jinja (#32053) 956591f45 ci : remove dead snakes (#32084) ccd360f06 llama : remove non-fused GDN and ubatch-dependent graph nodes (#32078) b326dd675 devops : add missing MKL runtime DLLs to the Windows SYCL release package (#32005) b8abe41a5 vendor: update cpp-httplib to 0.66.0 (#32049) 2e325dcd5 backend : re-instantiate CUDA graphs when the cluster dims of a kernel change (#31284) 1648d6411 test-fp8.cpp: use only the public ggml API (#32083) 432433637 ROCm/HIP: fixed BLAS workspace instead of rocBLAS allocating one per call (#31838) 71a6a114b quantize: fix imatrix GGUF filepath metadata leak (#32000) 7f24aacde cpu: accumulate flash attention V in F32 (#29810) af6b5243d webgpu: add PAD_REFLECT_1D support (#31877) d2b3eb9b6 vulkan: allow dequantized KV optimization with CM2 (#31526) eeb210479 vulkan: allocate the device fault vendor binary buffer (#32008) 03d6c824e tests : use only the public ggml API (#32077) adde94c11 ui: open the virtual keyboard in the chat editor on Android (#32060) 459ab1cdd cuda : use MMVF for BF16 matrices of few rows at small batch sizes (#31959) 77e2db8ca cuda : add sm70 FlashAttention config case for DKQ=256, DV=256, ncols=64 (#27997) 9d7b2d739 cuda : fattn mma head sizes 40 72 (#30285) 9e9619d03 CUDA: Add OCP FP8 support (#28413) f3b407ca1 sycl : fixed sycl CLAMP to actually support strided view (#31784) aa75ff323 chat : fix Muse Spark ignoring response_format json_schema (#32017) b5cb8f6f4 ui: keep a hidden sidebar from intercepting clicks again (#32035) 85822a7ab ui: update the README banner to the new llama.cpp cover and remove the legacy brand assets (#31789) 2b66bfbf7 ui: show audio in Discover for models whose mmproj has an audio encoder (#31837) 582eab718 ci : bump ty to 0.0.100 (#32030) 7c2653acc mtmd: fix string format, rename to mtmd_format (#32040) 7592b61d3 ui: move Models, MCP and Settings to a sidebar footer (#32022) cd4ed902e cuda : remove unused ids and type_src0 parameters (#32018) 202c1a229 llama : Add FP8 KV-cache support (#28392) e24d0b4c3 Merge pull request #32015 from ggml-org/gg/fix-sampling-output-aliasing 3e86ce2c2 llama : reuse backend scheduler when sched_params are compatible (#32009) 494d0f47b ggml : add MUL_MAT src1 precision tests and gating (#31896) 281e7dad7 ui: gather chats into named groups in the sidebar (#32012) 90e3841d9 ui: show settings save button even while focused (#32010) c75fba30c cuda : remove unused quantize_cuda_t typedef (#31884) 40f4f4ac9 graph: mark every sampling chain as an output to prevent aliasing 54cde7106 ui: expand the chat form into a large editor for longer messages (#31973) 7d6fd845e convert : mark Sarvam byte fallback tokens as BYTE (#32006) 4ed88da2e convert : workaround conversion for p-e-w/gpt-oss-20b-heretic-ara-v4 (#31860) ac931b5bc cuda : halve VMM pool reservation size and retry on cuMemAddressReserve failure (#30542) 15759e0ca CUDA: skip graph compat check when graph is being reused (#32004) a591cbe17 mtmd: share compute buffer among subcontext (#31945) 203b5f404 server: fix output lang for realtime api (#31982) 351025ec5 mtmd: add kimi K3.1 vision (#31980) 0d9eacadf ggml : fix FreeBSD Vulkan build by setting _XOPEN_SOURCE=700 (#31701) 17701e48d CUDA: move graph and concurrent stream state out of common.cuh (#32001) ac007054e ui: use a live-preview markdown editor for the chat form input (#31955) 940a30ee3 model : stop renaming the recurrent cache tensors in plamo2 (#31984) 45c8ea837 common/json-schema : fix allOf with plain types (e.g. integer) generating an empty object grammar (#31834) c90aa996b common : escape control characters in tool argument strings (#31937) a2c942dcf cann: fail early when checking if buffer type is supported (#31270) 18c5a95e2 sycl: Add device-side MoE scheduler (#30156) ae4893a09 vulkan: add tiled-transpose concat shader for Qwen models (#30591) 6122b566b ggml-rpc : fail gracefully on unreachable endpoints and failed handshakes (#30415) 3837bf0bc vulkan: Divide GROUP_NORM workgroup to 2D to avoid maxComputeWorkGroupCount limit (#31930) c1a4febe9 vulkan: Fix out of bounds reads/writes on MoltenVK (#31890) 78ffbaa18 vulkan: cap scalar flash attention at 8 rows for large heads on Qualcomm (#30829) 58e38bdf6 llama : apply the iGPU lazy-mode fallback only when mmap is off (#31957) d5bbcd1f3 ci: update oneAPI 2026.2, oneDNN 2026.1 and Level Zero SDK 1.34.0 (#31540) 8766c0273 sycl: un-gate oneDNN fused SDPA for Alchemist at head_dim=64 (#31574) 3b4d6ab3f chat : don't use lazy grammar when response_format is set with tools (#31956) cf6463ace hexagon: fix 64-bit FastRPC handle truncation on close (#30952) 711f9f4c1 server: responses api: support conversion of text.format.json_schema to completions (#26145) 26d52b0f9 server : use common_chat_input in realtime api (#31978) 08f945ffb server: add oai-compat /v1/audio/speech endpoint (#31763) 20c2e9665 mtmd: support gemma 4 input audio stream (#31965) 3d94ffe45 mtmd: add mtmd_helper_resampler (#31964) 0a94a3f28 ui : retry opening the sidebar when a demo imports its traces (#31952) 610f7ab40 ui: make the app feel native on phones (#31949) 5f490d7bf feat: external PLE file loading (#31185) 9d340ec34 Windows unbuffered model load (#26014) 359dc5858 server : functionality test for new spec dec (#31027) a4d4f2fd4 ggml-zdnn: accept non-zdnn tensors, invalidate compute tensors after every op (#31824) 1945793e0 server : fix response_format json_schema when schema is not nested in the json_schema wrapper (#28697) 0eebb6edd ggml : implement the backward pass for TRI (#30712) d628a98f0 tests : keep dummy ffn.act scales away from zero (#31940) a8fad21e9 mtp: read the target hidden states from a device buffer (#31927) b39680c41 ui: fold, preview and rearrange the sidebar's sections (#31936) 806b3c2be ggml-backend: don't upload inputs that are not read (#31419) 0c6dca2fa models : Attach Granite NVFP4 weight scales (#31449) fc0be8618 ci : Run gfx1151 ROCm job on Strix Halo Devlab runners (#31567) e01c9f187 metal : prevent out-of-bounds writes to the Metal FA output buffer (#31662) 4b46d9653 metal : make ggml_tri select instead of scaling by the comparison (#30711) f3178366b opencl : include climits (#31935) 2d99adb10 mtmd: Pocket TTS - add generic support for temperature set via API (#31655) cf92c1323 RPC: Improve loading times (#29413) 9f0d80cf4 ggml : support YaRN truncation config on openvino, sycl, vulkan, webgpu, rpc (#30802) a0ad695cb sycl: cache device-id (#31126) 6590d25a3 tests: bound the test_rope tolerance by the f32 angle resolution (#31874) 53eb66943 opencl: add f32 MUL_MAT_ID (#31783) d78f4255d ggml-zdnn: fix release binary not detecting backend lib (#31905) 7d7f02875 mtmd: refactoring: remove clip.h, consolidate code into mtmd (#31865) b65bf6921 cpu: enable SpacemiT IME2 MXFP4 dense repack (#31776) a3c7cf08f llamafile: impl bf16 and q8_0 vxe (#31866) 57d04820d ggml : update CUDA/Metal radix Top-K to make tied selections deterministic (#31084) 998200491 ggml-cuda : add d_state=64 fast path to SSM_SCAN kernel (#31818) cf34d3113 moe-cache : ensure CPU part does not wait for the cached experts (#31888) a691b2345 webgpu : add COUNT_EQUAL support (#31578) 2277e7529 ci: move runners to Ubuntu 26.04 (#31529) 12ab59798 ggml : switch to F32 src1 in the CPU backend (#31854) d6be06557 llama, mtmd: follow the reference precision of ggml_prec_set_acc calls (#31851) a2b301831 speculative : do not apply the MoE cache to the draft model (#31497) c19043f67 opencl: improve q8_0 and MoE small-n verification for spec (#31831) 372dc5ca1 common : sanitize invalid UTF-8 in PEG tool call arguments (#31861) f477f795c server : downgrade per-request proxy logs to trace (#31859) 76a67730f hexagon: support for automatic loading of libhtp without legacy env vars (#31781) 12e9084dc sycl: exact rounding in CPY to Q4_1, Q5_1 and Q8_0 (#31465) 818ac6b07 sycl: refuse device buffers that would exceed VRAM (#31285) e9a2a44fe hexagon: n_hvx and n_threads cleanup (#31857) a2654ec43 mtmd: move log-mel computation to cgraph (#31808) 9f0e5098a mtmd: support MageVL-Cowork model (#31819) e95426f16 llama_moe_cache: run missed experts on CPU, add tuning script (#30966) f7591a6a0 sycl: add f16 support for SWIGLU_OAI (#30857) 80d2c0c2c metal : fix GROUP_NORM for ne03 > 1 and non-contiguous inputs (#31852) 095668452 server: visit parent slots before child slots (#30779) cf659f240 webgpu : Implement group norm kernel (#31490) 193815264 vulkan: Add 'readonly' to decodeBuf declarations. (#31833) 5d7d4aec1 CUDA: fix shared memory bank conflicts in the lightning indexer (#31803) c0ab6e5c0 llama : add missing link to pthread (#31785) 31424b7cf opencl: add arange (#31829) fd27b2034 meta : raise the compute view limit for buffers with few tensors (#31447) ca89968a2 hexagon: fuse post-norm residual add into RMS_NORM_MUL (#31822) bcb015287 model: add Kimi-K3.1-Pro text model (#31779) 669204ee2 hexagon: conv_2d optimizations and support for conv_3d (#31814) 954de248a docker: add explicit /usr/local/nvidia search path (#31806) 6c01da627 cli : fix crash on invalid UTF-8 in prompt or files (#31563) 8820b088d tests : qualify isnan and signbit with std:: (#31791) 9a1d03c7c tests : override hardware_concurrency in CI (#31797) 18ab62a8e cmake : generate RC file directly without win-app.rc.in (#31718) 52c9bad94 opencl: support repeat for f16, i32, i16, bf16 (#31782) 5d3ed2ae7 webgpu: add batched tensor API (get_tensor_n and set_tensor_n) (#31473) 80088fa60 vulkan: detect ReBAR support, avoid BAR allocations if unavailable (#31658) c4fb57969 moe-cache : LRFU/MQ-like hybrid cache, ClassLRU (#31233) 3ed7284f2 ci : bump CUDA 13 release builds to 13.5 (#31788) d9cf1870b vendor: update BoringSSL to 0.20261105.0 (#31757) 240d68fcf CUDA: fix Maxwell build (#31787) bbc2e4b65 common : restrict HF token to huggingface.co hosts (#31730) 4b77dee05 opencl: add q6_k multi column tiled GEMV for speculative verification and refactoring (#31758) 67107ac34 hexagon: Implement set_tensor_n and improve repacking/tiling robustness (#31744) e14a77971 mtmd: enable cmake unity build (#31760) d38bea342 tests : drop u8path() (#31755) 186eab1cb ci: fix build-cmake-pkg ccache (#31705) 6023692b1 llama: fix kv shifting, add tests, add per-model build_rope (#31683) b1c902f02 llama, mtmd: (chatterbox) add use_classifier_free_guidance as gguf metadata (#31698) e8df20d23 models : recursive fix (#31749) fecdc81ee model : add Ripple (ricursiveai) (#31692) 45b6604b2 model: Add support for the xing4_0 (#31589) e30ad80fd ui: prevent a hidden sidebar from intercepting clicks (#31712) e0f96a1a1 llama: refactoring: move model to class member of llm_graph_context (#31733) 468fde90e server : simplify windows MCP command resolution (#31728) 1416f76aa vendor : update cpp-httplib to 0.65.0 (#31731) 10d4846a5 ggml-cuda: sanitize post-bias NaNs in MoE selection (#31005) 61d447234 vulkan: Implement test-fusion interface (#30769) 016c1d914 CUDA: use highest compiled arch in get_mmvq_mmid_max_batch (#31679) da3416d80 vulkan: RDNA3/4 Large-Tile Split-K Optimization (#30280) cc78da97c gguf-py : external chat templates have precedence over tokenizer_config (#31717) ea40a3a1e metal : fuse DeepSeek-V4/V4.1 and GLM MoE routing (#31615) 15e50990b CUDA/HIP: matrix-core (MFMA) lightning indexer for CDNA2 (gfx90a) (#29050) 69ee2ecc0 CUDA: linearize MoE data layout for MMQ/MMF (#31047) 2bc46baa9 vulkan: FWHT kernels for block widths above 512 (#29772) 3b83d12b0 vulkan : check combined K+V f16 scratch size in FA dequant path (#30487) e152332a1 llama-server: Use reference-counted checkpoints instead of deep-copies (#30137) 1fa7f1c48 ggml : support truncate=false for YaRN (#30206) 9c9074e14 conversion : resolve quantized_layers from source tensor names (#31611) f7b41cca4 metal : use mul_mm for 32 <= K < 64 (#30982) c55b09488 server : reject non-object session updates (#31714) 2c192444c fit : account for skipped NextN layers when setting n_gpu_layers (#31569) 553d2c2e7 webgpu : add MEAN support (#31591) 040ce199a graph: store the MTP conv snapshots of every recurrent model in one set_rows (#31660) d4c345d6c ci : remove unused model generation (#31412) 2ada83079 common : drop u8path() and use generic_string() (#31681) 5a081c0a7 server : drop u8path() and use generic_string() (#31682) 7e42b95f2 rpc : align code with common.cpp (#31684) 3259ab6e5 vulkan: Re-enable large coopmat matmul configuration on AMD proprietary driver 26.10.1+ (#31587) a01dff908 tests: fix CI flakes of TOPK_MOE caused by near ties (#31426) 44f91a02b metal: keep fusable node groups inside one command buffer (#31688) 330f0189c qwen2 : add nvfp4 scales to build_ffn (#31657) 7d516e25e opencl: enable f32 and q5_k multi column GEMV by default for Adreno X2 (#31694) 8bed7cfb9 opencl: support non-contig sigmoid (#31687) 8415bd7bb opencl: add DSV4_HC_PRE and DSV4_HC_POST (#31623) 5ce06fc71 Title: CUDA: fuse NVFP4 gate/up matmuls and GLU in MMQ (#31296) e56f006ed llama : fix on-device state copy sizes for quantized KV caches (#30541) 3de52ac38 ci: run self hosted cpu jobs on hf or github (#31439) 6f1e65437 server : realtime: send an error event instead of crashing on exceptions (#31632) 9e2a04a43 opencl: improve decode perf with FA (#31633) c6a0ff11a ui: open actions on right-click for chats, tabs, messages and models (#31545) 5851a8e50 model, mtmd: drama-3 tts support (#31613) efbf6e80a ui: show text and page thumbnails for PDF attachments, stop endless conversion retries (#31468) b393d73bb webgpu: return maxBufferSize from get_max_size and avoid whole-buffer bindings in memset/clear (#31348) 00df7c22d sycl : fix docker issue (#31605) 034e2535a gguf-py : validate every tokenizer merge pair, not only the first (#31631) 52e53a76d common : delete common_get_path_from_env() (#31444)