llama.cpp Releases
609 releases of ggml-org/llama.cpp
- b9988
A build system improvement to ensure tests use the shipped headers from the project rather than potentially system-installed ones, ensuring consistency and avoiding conflicts during compilation.
Jul 13, 2026
- b9987
This release adds a new accessor function to the GGUF library to retrieve tensor shapes, improving the ability to inspect model metadata programmatically. It also refactors the accessor interface by removing the deprecated `n_dims` accessor.
Jul 13, 2026
- b9986
A bug fix for the chat interface was implemented to prevent reasoning content from leaking when using force-opened bare templates, specifically addressing a regression with Nemotron Nano v2.
Jul 13, 2026
- b9985
This release introduces fused top-k Mixture-of-Experts (MoE) support for the SYCL backend, optimizing performance for specific hardware configurations. It also includes various binary builds across different platforms (macOS, Linux, Windows, Android) and architectures.
Jul 13, 2026
- b9984
Adds support for Q2_K quantization format to the SYCL (Intel oneAPI) DMMV (Dynamic Multi-Vector) reordering path, improving performance and compatibility for Intel hardware.
Jul 13, 2026
- b9982
Fixes a bug where per-request overrides for `reasoning_budget_tokens` and `reasoning_budget_message` in chat completions were ignored in favor of server defaults, breaking support for advanced reasoning models like Qwen3.
Jul 13, 2026
- b9981
Updates the embedded HTTP server library (cpp-httplib) to version 0.50.1, incorporating the latest fixes and improvements from the dependency.
Jul 13, 2026
- b9980
Fixes the `/v1/models` endpoint to correctly report input modalities, ensuring that models with multimodal projection explicitly disabled via `--no-mmproj-auto` are not falsely advertised as multimodal.
Jul 13, 2026
- b9979
Fixes a silent prompt truncation bug where prompts containing embedded NUL bytes were cut off at the tokenizer boundary, causing subsequent messages and markers to be dropped.
Jul 12, 2026
- b9978
Improves checkpoint management in the server by evicting checkpoints that are too close in step count, helping to optimize storage usage and prevent accumulation of redundant files.
Jul 12, 2026
- b9977
Fixes a bug where image blocks in tool results were silently dropped during the Anthropic-to-OpenAI conversion process, breaking multimodal tool outputs.
Jul 12, 2026
- b9976
Fixes the conditional logic for displaying an error message regarding unsupported tensor split modes on specific architectures, ensuring users receive appropriate feedback.
Jul 12, 2026
- b9975
Fixes a GGUF validation issue by rejecting empty metadata keys to prevent potential loading errors. Includes standard binary builds for macOS, Linux, Windows, Android, and openEuler with support for various hardware backends.
Jul 12, 2026
- b9974
Prevents crashes in CUDA builds when querying memory on devices with insufficient memory. This allows the application to gracefully handle low-memory devices and explicitly disabled devices via the `-dev none` flag.
Jul 12, 2026
- b9973
Fixes the cache clearing logic for DeepSeek V4 to clear cache only for specific sequences instead of the entire context, improving efficiency.
Jul 12, 2026
- b9972
Enhances the server API by enabling streaming for the `exec_shell_command` tool, allowing real-time output from shell executions.
Jul 12, 2026
- b9971
Refactors the server stream functionality to improve thread safety and code organization by removing a non-thread-safe stop call and relocating server components.
Jul 12, 2026
- b9970
Introduces a new `GGML_OP_LIGHTNING_INDEXER` operation to implement the lightning indexer for DeepSeek V3.2 and V4 models, along with comprehensive tests and RPC updates.
Jul 12, 2026
- b9969
Fixed a critical crash bug in llama-cli for longer prompts by updating Vulkan matmul routing for Adreno GPUs to use medium tiles and fixing shared memory issues.
Jul 12, 2026
- b9968
Added optimized int8 dense matrix multiplication and MoE (Mixture of Experts) prefill operations specifically for Adreno GPUs via the OpenCL backend.
Jul 12, 2026