Open-Source Models & Communities — September 22, 2026

Findings

1. [1] 2026-09-22 b11112

   server: support input_image in function_call_output (#20663) (#22575) server: support input_image in function_call_output (#20663) server: fix if statement spacing server: avoid repeated type lookup Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/49330467 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan)…

2. [2] 2026-09-22 b11111

   vulkan: add Intel Xe flash attention optimization kernels (2/3, Xe-LPG Plus/Xe2/Xe3) (#24406) vulkan : Intel FA kernel optimization for split k path vulkan : Host code update for Intel split k FA kernel path selection, fix A770 Linux op test failures vulkan : use symmetric coopMatMulAdd() in flash_attn_decode_phase_1 shader to resolve test op failre on A770 Linux with 26.2.3 mesa…

3. [3] 2026-09-22 b11110

   mtmd: add various sanity checks (#29276) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/49308522 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (CUDA 12) – CUDA 12.8 libraries Ubuntu x64 (CUDA 13) – CUDA 13.4 libraries Ubuntu…

4. [4] 2026-09-22 b11109

   metal : gate mul_mm_id src1 rescale behind ggml_prec (#29029) metal : gate mul_mm_id src1 rescale behind ggml_prec Assisted-by: Claude Fable 5.1 ggml-webgpu: reject MUL_MAT_ID when src1 precision is F32 cuda/vulkan: reject MUL_MAT_ID in supports_op when src1 prec is F32 fix supports_op to return false for failing backends when the specified src1 precision is f32 Assisted-by: Claude Fable 5.1 Co-authored-by: yomaytk…

5. [5] 2026-09-22 b11108

   ggml : IQ1_M build prefix sums once per block (#28706) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/49294443 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (CUDA 12) – CUDA 12.8 libraries Ubuntu x64 (CUDA 13) -…

6. [6] 2026-09-22 b11105

   jinja: use const for statement::execute and ::visit (#29271) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/49279983 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (CUDA 12) – CUDA 12.8 libraries Ubuntu x64 (CUDA 13) – CUDA 13.4…

7. [7] 2026-09-22 b11104

   server: Add support for binding to multiple addresses (#28690) Add support for binding llama-server to multiple addresses Assisted-by: Codex remove redundant thread handler make it clear about overlapping addr reject –port 0 with multiple tcp addr improve arg handler nits fix test nits 2 nits nits 2 Co-authored-by: Xuan Son Nguyen son@huggingface.co Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/49267001 macOS/iOS: macOS Apple Silicon…

8. [8] 2026-09-22 b11103

   spec : support DFlash for HunyuanOCR (#28890) model : add DFlash layer-input taps for HunyuanVL DFlash speculative decoding needs the target graph to expose the residual stream entering each layer (res->t_layer_inp[il]) – the draft model reads those tensors to build its cross-context. Qwen3 and the other DFlash-capable targets register them, but the Hunyuan graphs do not, so serving a DFlash…

9. [9] 2026-09-22 b11102

   convert: add MiMo-V2.6 support (#29257) convert: add MiMo-V2.6 support Hoist the K3 mxfp4 conversion repack into base.py so it can be reused Remove decoder from mmproj convert Update conversion/mimo.py fix: use autoparser Co-authored-by: Sigbjørn Skjæret sigbjorn.skjaeret@huggingface.co Co-authored-by: Piotr Wilkin piotr.wilkin@syndatis.com Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/49229124 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS…

10. [10] 2026-09-22 b11101

    cmake : allow repeated find_package calls for llama (#29228) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/49213556 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (CUDA 12) – CUDA 12.8 libraries Ubuntu x64 (CUDA 13) – CUDA…

11. [11] 2026-09-22 v0.30.0

    v0.30.0 Highlights This release features 762 commits from 315 contributors (104 new)! New models: DeepSeek-V4.1-Flash (#56214, #56228, #56208) with the whole KV stored in MXFP8 through the FlashMLA V4.1 record on SM100 (#56893), DeepGEMM Mega-mHC (#56962), and async Engram prefetch… New models: DeepSeek-V4.1-Flash (#56214, #56228, #56208), DeepSeek-V4-Flash-Vision-Exp (#54566) with ROCm support (#55107) and LoRA (#55897), GLM-5.3-Flash (#53906), K2-Horizon with its reasoning and tool parsers (#55063), Cohere Compass (#54774, #55949), Bailing V3 VL with MTP (#55921), Nanbeige4.2 via the Transformers backend… Correctness: Gemma 4 / Gemma 3n fine-tuned checkpoints without KV projections on shared layers (#54917, #55864), Gemma4 video preprocessing without image kwargs (#56652) and unified audio profiling (#56721), MiniCPM-V 4.6 loading (#49417) and video embeds (#54633), InternVL video parser (#55779),… Engine Core Model Runner V2: DBO in eager mode (#50945) and with FULL CUDA graphs (#51700), MTP (#46994) and EAGLE3-style drafts (#50514) under pipeline parallelism, adaptive verification for all draft-model speculators (#52228), gc frozen during capture (#54646), GPU-compacted sampling masks… Hardware & Performance NVIDIA: FlashInfer PCIe IPC all-reduce (#53576), DeepGEMM pinned to the vLLM fork 2.8.0 with SM120 and SM90 paged-MQA ports (#56876), public CUDA 13.4 Rubin build path (#54640, #56545), SM12x blockwise FP8 CTA raster swizzle for GB10/DGX Spark… Context parallelism: PCP+DCP on sparse-MLA models (#56157), PCP with single-module MTP and replicated DSpark (#56107), PCP decode-only FULL CUDA graphs (#53867), FlashMLASparse MTP at CP interleave > 1 (#56722), symmetric DCP disaggregation for hybrid Mamba models (#55531), DCP empty-shard LSE…

12. [12] 2026-09-22 How UK AISI and EvalEval Are Making Benchmark Results Reproducible

13. [13] 2026-09-22 Transformers now runs llama.cpp quants

Sources

1. [1] b11112
2. [2] b11111
3. [3] b11110
4. [4] b11109
5. [5] b11108
6. [6] b11105
7. [7] b11104
8. [8] b11103
9. [9] b11102
10. [10] b11101
11. [11] v0.30.0
12. [12] How UK AISI and EvalEval Are Making Benchmark Results Reproducible
13. [13] Transformers now runs llama.cpp quants