Back to Freedom.Tech
Project AI

llama.cpp

C/C++ inference engine that powers most local LLM runners. Tagged releases.

126k Stars
22k Forks
98
Releases tracked
GitHub releases
Source type
1 day
Avg cadence
May 2026
First tracked
today
Latest

Release history

AI release Aug 27, 2026

llama.cpp b10644

Adds nanbeige4.2-3B model support

AI release Aug 26, 2026

llama.cpp b10636

Ui: disable the npm UI build (LLAMA BUILD UI=OFF)

AI release Aug 25, 2026

llama.cpp b10625

Scopes Qwen3-coder chat workarounds

AI release Aug 25, 2026

llama.cpp 0.3.0

Llama.cpp 0.3.0 adds dots3-note multimodal support, tensor-split for DeepSeek 4, multi-token prediction for GLM-4.5-Air, and WebP image decoding

AI release Aug 24, 2026

llama.cpp b10610

Metal flash-attention tuning: 53 new f16 instantiations, per-device dispatch tables

AI release Aug 23, 2026

llama.cpp b10603

GLM-4.5-Air MTP support added to llama.cpp for improved model inference efficiency

AI release Aug 22, 2026

llama.cpp b10588

Fixes clang LTO build issue

AI release Aug 21, 2026

llama.cpp b10568

CI switches to LLVM OpenMP on Windows; Drops non-redist debug build

AI release Aug 21, 2026

llama.cpp 0.2.0

Llama.cpp 0.2.0 syncs ggml to 0.21.0, adds backend optimizations for CUDA/Metal/SYCL/OpenCL, and improves quantization kernel performance across hardware

AI release Aug 20, 2026

llama.cpp b10505

Adds dedup-cache-models preset option to server

AI release Aug 18, 2026

llama.cpp b10486

Fixes LFM2 image tiling threshold; Cross-platform refactor

AI release Aug 18, 2026

llama.cpp 0.1.2

Llama.cpp synced GGML 0.20.2, added integer tokenizer scores, improved CUDA performance on DGX, and fixed xcframework builds

AI release Aug 17, 2026

llama.cpp b10470

CI; Explicitly pushes release tag before creating GitHub release

AI release Aug 16, 2026

llama.cpp b10452

Removes some ggml concat calls; Model-layer refactor

AI release Aug 15, 2026

llama.cpp b10447

Adds Kimi-K3 text-model support with hybrid KDA and MLA attention

AI release Aug 14, 2026

llama.cpp b10434

Llama.cpp now passes reasoning_effort parameter through chat templates, enabling fine-grained control over model reasoning depth in inference

AI release Aug 13, 2026

llama.cpp b10419

OpenVINO backend gains Qwen3.5 support, memory optimizations for GPU inference, and fixes for stateful RoPE accuracy and recurrent state handling

AI release Aug 12, 2026

llama.cpp b10369

Build b10375 tightens bare function parsing for Qwen models

AI release Aug 11, 2026

llama.cpp b10359

Switches ROCm target to 7.14, the first production release using TheRock build system

AI release Aug 10, 2026

llama.cpp b10344

Added MTP (Multi-Token Prediction) support for Nemotron models, enabling faster inference for this architecture

AI release Aug 8, 2026

llama.cpp b10329

Corrects server info endpoint to report the isolate working directory when a tools runtime is configured

AI release Aug 7, 2026

llama.cpp b10326

TTS timing now accounts for full vocoder pass instead of single trailing window

AI release Aug 6, 2026

llama.cpp b10290

AMD ROCm CI support for gfx1151; No inference changes

AI release Aug 5, 2026

llama.cpp b10289

Fixed directory traversal vulnerabilities in file search, improved path handling on Windows with UTF-8 conversion, and added cache expiration for UI picker searches

AI release Aug 4, 2026

llama.cpp b10271

Adds a per-conversation working directory with file picker; Agents can now treat path-like queries as directory navigation

AI release Aug 3, 2026

llama.cpp b10236

Implements DSv4 Lightning Indexer and f16 Lightning Indexer for Metal; Affects 128-dimensional 64-head inputs on Apple silicon

AI release Aug 2, 2026

llama.cpp b10232

WebGPU backend adds f16 repeat support; MacOS KleidiAI build disabled

AI release Aug 1, 2026

llama.cpp b10219

Llama.cpp now preserves reasoning content in chat history, enabling multi-turn conversations to leverage previous model reasoning with --reasoning-preserve flag

AI release Jul 31, 2026

llama.cpp b10212

Adds driver-version check for Intel GPUs on Windows to avoid crashes; No action unless running that config

AI release Jul 30, 2026

llama.cpp b10195

Tests alternative convolution layout; Extends layout checks for conv2d kernel

AI release Jul 29, 2026

llama.cpp b10182

Adds NextN and MTP speculative decoding for GLM-5.2; GLM DSA support

AI release Jul 28, 2026

llama.cpp b10173

Added Laguna-S-2.1 model support to llama.cpp for local inference

AI release Jul 27, 2026

llama.cpp b10153

MiMo-V2.5 RVQ-based audio input support added

AI release Jul 24, 2026

llama.cpp b10105

Refactors mlock, mmap, and direct I/O into unified load-mode argument

AI release Jul 23, 2026

llama.cpp b10099

CUDA NVFP4 W4A4 activation quantization improvements, fused per-channel amax and quantization kernels

AI release Jul 22, 2026

llama.cpp b10090

Adds depthwise conv2d kernel for WebGPU backend, ported from Vulkan

AI release Jul 21, 2026

llama.cpp b10078

Vulkan queue refactoring enables driver-side synchronization on supported hardware, reducing host-side mutex contention for GPU inference

AI release Jul 18, 2026

llama.cpp b10067

K/V cache rotation fix for DFlash quantization workflows

AI release Jul 17, 2026

llama.cpp b10064

OpenCL kernel gemm_moe_q6_k_f32_ns now loaded from binary kernel library. Binaries for macOS, iOS, Linux, and Windows published

AI release Jul 16, 2026

llama.cpp b10051

MacOS, iOS, and Linux build artifacts now included in release notes

AI release Jul 15, 2026

llama.cpp b10032

Fixes OpenCL buffer creation for CL 2.x; MacOS arm64 and Linux CPU builds included

AI release Jul 14, 2026

llama.cpp b10012

Adds contiguity checks for inner tensor dimensions

AI release Jul 13, 2026

llama.cpp b9992

Corrects server endpoint to honour per-request reasoning-budget tokens in chat completions

AI release Jul 12, 2026

llama.cpp b9977

OpenCL int8 dp4 dense and MoE GEMM optimization for Adreno GPUs

AI release Jul 11, 2026

llama.cpp b9959

Llama.cpp synced ggml changes, updating core ML infrastructure for model inference

AI release Jul 10, 2026

llama.cpp b9952

Llama.cpp optimizes DeepSeek V4 inference with f16 KQ masks, removes unnecessary attention bias, and eliminates redundant cache repeats for faster execution

AI release Jul 9, 2026

llama.cpp b9945

Hexagon: tiles wide rows in unary ops to avoid VTCM overflow, replaces divs with fastdiv, adds VTCM layout and host-computed offsets

AI release Jul 8, 2026

llama.cpp b9910

Fuses MMVQ post-scale and batch-size-1 path for NVFP4

AI release Jul 6, 2026

llama.cpp b9890

Adds Metal col2im 1d op support for f32, f16, bf16

AI release Jul 5, 2026

llama.cpp b9878

Fixes stale tensor-split metadata for GQA attention on draft models

AI release Jul 4, 2026

llama.cpp b9873

Adds guard to prevent dereferencing unallocated K/V rotation buffers; Defensive fix for edge-case crashes

AI release Jul 3, 2026

llama.cpp b9866

Trims messages sent to StepFun parser, fixes long reasoning loops

AI release Jul 2, 2026

llama.cpp b9860

Adds llama_model_ftype_name function to expose quantization type as string

AI release Jul 1, 2026

llama.cpp b9857

Uses Hugging Face primary split as model path; Fixes issue 25181

AI release Jun 30, 2026

llama.cpp b9849

Registers attention normalization tensor for Qwen3; Fixes input assignment in layer loop

AI release Jun 29, 2026

llama.cpp b9839

Adds DeepSeek V4 conversion support including pro model, rope fixes, and chat template

AI release Jun 28, 2026

llama.cpp b9835

Fixes UI stop and reasoning skip in single-model mode

AI release Jun 27, 2026

llama.cpp b9825

Vulkan backend fix for zero-input step operator handling in llama.cpp inference

AI release Jun 26, 2026

llama.cpp b9821

Adds support for version, licenses, and help flags

AI release Jun 25, 2026

llama.cpp b9788

Fixes quantizing MoE models with mtp

AI release Jun 24, 2026

llama.cpp b9776

Adds LFM2.5-ColBERT-350M and LFM2.5-Embedding-350M support; No impact unless loading those weights

AI release Jun 23, 2026

llama.cpp b9767

Improves WebGPU mat-vec path for small-batch decoding

AI release Jun 22, 2026

llama.cpp b9760

Relevant if running the server component and permitting file edits, appends at end of file no longer trigger heap overflow

AI release Jun 21, 2026

llama.cpp b9752

Server batch construction refactored for more careful slot handling and improved abort logic

AI release Jun 20, 2026

llama.cpp b9739

WebGPU adapter toggles for F16 on Vulkan and NVIDIA. Relevant only if running those backends

AI release Jun 19, 2026

llama.cpp b9732

Server context-shift fix for non-bound n_discard value. Upgrade if running server mode with ctx shifting

AI release Jun 18, 2026

llama.cpp b9694

Fixed Windows x64 OpenVINO release build link to resolve CI pipeline failures

AI release Jun 17, 2026

llama.cpp b9690

Optional USM system allocations for buffers over 1GB if the device supports it

AI release Jun 16, 2026

llama.cpp b9670

Adds Vulkan support for gated delta net with S v=16

AI release Jun 15, 2026

llama.cpp b9642

CUDA operators note F32/F16 restriction on GGML OP REPEAT; KleidiAI macOS build remains disabled

AI release Jun 14, 2026

llama.cpp b9628

Adds Cohere2MoE vocab for TINY AYA. No impact unless loading those weights

AI release Jun 12, 2026

llama.cpp b9611

Llama.cpp refactored build dependencies to reduce coupling between fit and llama-ext headers

AI release Jun 11, 2026

llama.cpp b9594

Refactors normalizer flags into options struct and adds strip-accents support. No model or inference changes

AI release Jun 10, 2026

llama.cpp b9589

CUDA: Fix ssm_scan_f32 data-races missing syncthreads calls to prevent shared memory corruption

AI release Jun 9, 2026

llama.cpp b9585

Operators running llama.cpp with plamo2 models should upgrade. Fixes an attention key-value length regression introduced in a prior build

AI release Jun 8, 2026

llama.cpp b9568

Fixes im2col 1D case for audio models. Relevant only if running audio inference

AI release Jun 7, 2026

llama.cpp b9553

Sampler name matching now case-insensitive and accepts alternative names like top-k alongside canonical top_k format

AI release Jun 6, 2026

llama.cpp b9538

Qwen3.5 video support added; No impact unless loading those weights

AI release Jun 5, 2026

llama.cpp b9533

Build system fix resolves compilation failures in llama.cpp, enabling reliable local LLM inference on consumer hardware

AI release Jun 4, 2026

llama.cpp b9498

WebGPU FlashAttention refactored with standardized quantization support, relevant if running WebGPU backend

AI release Jun 3, 2026

llama.cpp b9486

Adds support for qwen3 SSM architectures

AI release Jun 2, 2026

llama.cpp b9473

SWA checkpoint optimization reduces KV cache storage masked cells, improving memory efficiency for inference

AI release May 31, 2026

llama.cpp b9444

Adds tokenizer support for jina-embeddings-v2-base-zh. No impact unless loading those specific embedding weights

AI release May 30, 2026

llama.cpp b9434

Fixed tensor parallelism granularity for Qwen 3.5/3.6 models across multiple GPUs

AI release May 29, 2026

llama.cpp b9414

Adds DeepSeekOCR 2 support with multi-tile dynamic resolution

AI release May 27, 2026

llama.cpp b9368

Mesa users on Intel BMG should see a 4.8 percent tg128 improvement for Qwen3.5-9B BF16 workloads

AI release May 26, 2026

llama.cpp b9334

CUDA users should upgrade for the PDL sync fix; MacOS KleidiAI artifacts now shipping

AI release May 24, 2026

llama.cpp b9305

Fixes UI build and adds macOS KleidiAI artifacts. No impact unless building the UI or deploying on Apple Silicon with KleidiAI

AI release May 23, 2026

llama.cpp b9297

Adds NVFP4 MTP scale tensors and links Qwen3.5 MTP tensors. No action unless running Qwen3.5 or NVFP4 hardware

AI release May 22, 2026

llama.cpp b9289

SYCL backend now supports gated_delta_net operations with K 1 for improved inference performance on Intel GPUs

AI release May 21, 2026

llama.cpp b9258

Merges HunyuanOCR into HunyuanVL vision path and restores bilinear sampler precision for OCR tasks

AI release May 20, 2026

llama.cpp b9245

RDNA3 GPU kernels for Q6_K quantized inference now use tuned warp configurations for better performance

AI release May 19, 2026

llama.cpp b9222

Adds Hexagon TRI op support; Matters only if targeting Qualcomm HTP acceleration

AI release May 18, 2026

llama.cpp b9216

Llama.cpp UI refactored models store, MCP service, and debug logging behind environment flags to reduce noise and improve performance

AI release May 17, 2026

llama.cpp b9198

Vulkan build now checks for SPIRV-Headers availability upfront to catch configuration issues before compilation starts

AI release May 16, 2026

llama.cpp b9174

Moves web UI output directory and renames CMake variables. Existing build scripts may need path adjustments

AI release May 14, 2026

llama.cpp b9156

Adds Qwen3.5 tokenizer support. No impact unless loading Qwen3.5 weights

AI release May 13, 2026

llama.cpp b9129

Llama.cpp ZenDNN backend now adaptively falls back to CPU for small batch sizes, improving efficiency via runtime control