RepoJournal
Local LLMs

Local LLMs

Ollama, llama.cpp, vLLM, and SGLang - running and serving open models locally

Keep up with Local LLMs in about 3 minutes a day: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

One email a day. Unsubscribe in one click.

Pick a date

  1. Thu 23 jul
  2. Fri 24 jul
  3. Sat 25 jul
  4. Sun 26 jul
  5. Mon 27 jul
  6. Tue 28 jul
  7. Wed 29 jul
  8. Thu 30 jul
  9. Fri 31 jul
  10. Sat 1 aug
  11. Sun 2 aug
  12. Mon 3 aug
  13. Tue 4 aug
  14. Wed 5 aug
  15. Thu 6 aug
  16. Fri 7 aug
  17. Sat 8 aug
  18. Sun 9 aug
  19. Today 10 aug
Topics: Local LLMs AI / ML Full archive →

The Wire · Showcase

OLLAMA CUTS MODEL CONFIG BLOAT 100KB+ PER MODEL, LLAMA.CPP SHIPS TOOL ISOLATION AND CUDA FUSION GAINS

By RepoJournal · Filed · About Local LLMs

github-actions[bot] hnyls2002 Daniel Hiltgen Isotr0py YazhiGao grafail ylsun
11 people shipped this
github-actions[bot] @github-actions[bot] 4 cited hnyls2002 @hnyls2002 2 cited Daniel Hiltgen @dhiltgen 1 cited Isotr0py @Isotr0py 1 cited YazhiGao @YazhiGao 1 cited grafail @grafail ylsun @xiaopusun Nick Hill @njhill aarushjain29 @aarushjain29 cherryblo @cherryblo xdtbynd @xdtbynd

Ollama eliminated redundant manifest data ballooning config blobs on safetensors models, while llama.cpp landed docker-backed tool isolation and fused three GPU kernels into one.

Ollama's manifest layer deduplication [1] strips unused rootfs.diff_ids that were copying every layer digest into the config blob, cutting config size by over 100KB on per-tensor safetensors models without changing model identity. Llama.cpp shipped four releases in 24 hours rolling out production-grade tool isolation via docker [2], proper working directory reporting for isolated runtimes [3], and UI controls that only surface when tools actually need them [4]. On the GPU side, a new CUDA kernel fuses rms_norm plus mul plus rope operations, combining existing fusions and driving 1% throughput gain on affected models [5]. VLLm migrated bitsandbytes support to an out-of-tree plugin [6], trimming core dependencies while keeping BNB models working via simple package install. SGLang hardened KV cache freeing against aliasing bugs [7], unified scheduler write-after-read barriers across speculat and non-speculative paths [8], and cut 5.5 GPU-hours from nightly CI by removing redundant test registrations [9].

One email a day. Unsubscribe in one click.

Action items

References

  1. [1] manifests: remove OCI rootfs from the model config (#17619) ollama/ollama
  2. [2] b10328 ↗ ggml-org/llama.cpp
  3. [3] b10331 ↗ ggml-org/llama.cpp
  4. [4] b10329 ↗ ggml-org/llama.cpp
  5. [5] b10330 ↗ ggml-org/llama.cpp
  6. [6] [Migration] Migrate bitsandbytes support to OOT plugin ↗ vllm-project/vllm
  7. [7] [Bugfix] Fix batched KV free aliasing ↗ sgl-project/sglang
  8. [8] [Scheduler] Unify WAR read-done gating behind shared-read boundary declarations ↗ sgl-project/sglang
  9. [9] [CI] Trim redundant nightly test registrations ↗ sgl-project/sglang

Quick answers

What shipped in Local LLMs on August 9, 2026?
Ollama eliminated redundant manifest data ballooning config blobs on safetensors models, while llama.cpp landed docker-backed tool isolation and fused three GPU kernels into one. In total, 49 commits, 49 pull requests, and 4 releases landed.
Who contributed to Local LLMs on August 9, 2026?
11 developers shipped this update, including Daniel Hiltgen, github-actions[bot], grafail, Isotr0py, ylsun, Nick Hill, aarushjain29, and YazhiGao, and 3 more.
What were the notable Local LLMs updates?
manifests: remove OCI rootfs from the model config (#17619), b10328, and b10331.

More from Local LLMs

Local LLMs Local LLMs

OLLAMA SHIPS VISION SUPPORT ACROSS MLX AND QWEN3.5 AS LLAMA.CPP FIXES DISPATCH BUGS

Ollama's MLX runner now processes images natively, preserving prefix caching and speculative decoding while rejecting unsupported requests with clear errors.

jessegross Daniel Hiltgen github-actions[bot] allozaur +9
llms 157 shipped
Local LLMs Local LLMs

OLLAMA SHIPS BLOCK-DIFFUSION SPECULATIVE DECODING, SGLANG LAUNCHES KIMI K3 DAY-0

Draft models just got faster: Ollama merged DFlash block-diffusion speculative decoding that proposes entire token sequences in one forward pass, while SGLang shipped production-ready support for Kimi K3, a 2.8T-parameter multimodal LatentMoE that routes through a latent space with 896 experts.

jessegross ParthSareen hoyyeva dhiltgen +12
llms 263 shipped
Local LLMs Local LLMs

LLAMA.CPP HARDENS VULKAN STACK, VLLM CUTS ATTENTION LATENCY, SGLANG SCALES MLA

Vulkan submission batching and AMD ROCm CI fixes land in llama.cpp as vLLM optimizes attention kernels and sglang adds per-request MLA scheduling.

ngxson 0cc4m Jim Wu arthw +9
llms 231 shipped
Local LLMs Local LLMs

OLLAMA FIXES KV CACHE CORRUPTION, LLAMA.CPP HARDENS FILE SEARCH, VLLM KILLS MOE LEGACY CODE

Ollama patched a critical cache corruption bug that corrupts lazy snapshots during multi-request prefills, while llama.cpp hardened Windows junction traversal and vLLM removed 14 hardcoded MoE shards across quantization methods.

jessegross Daniel Hiltgen Xuan-Son Nguyen Niklas Wenzel +10
llms 240 shipped

Daily updates, in your inbox

Follow Local LLMs

Keep up with Local LLMs in about 3 minutes a day: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

or

One email a day. Unsubscribe in one click. Read a past issue →

Elsewhere on the wire

AI Agents AI Agents

LANGCHAIN PATCHES PYDANTIC 2.14 BREAKS, CLINE KILLS YOLO MODE

LangChain shipped critical compatibility fixes for Pydantic 2.14 serialization bugs while Cline removed a cosmetic toggle that masked real approval confusion.

langchain-oss-model-profiles[bot] ccurme langchain-oss-model-profiles saoudrizwan +1
ai-agents 10 shipped
CachyOS @CachyOS

CACHYOS LINUX 7.2-RC7 SHIPS WITH WIFI AND USB FIXES, SETTINGS 1.4.0 REMOVES NVIDIA DEPRECATED FLAG

CachyOS Linux 7.2-rc7-1 rebased upstream with critical fixes for MT76 wifi devices and USB initialization, while the settings package ditches legacy NVIDIA parameters.

Eric Naim ptr1337 Dharmik-Lathiya Vladislav Nepogodin +2
infra 73 shipped
Elixir & Phoenix Elixir & Phoenix

ELIXIR FIXES UTF-8 CORRUPTION IN MACRO.TO_STRING, ECTO CUTS COMPILE DEPS

A critical fix landed for Elixir's string handling that was silently corrupting UTF-8 sequences when converting C1 control characters, while Ecto shipped a change that trims unnecessary compile-time dependencies from schema derivations.

mnnekrashevich woylie
elixir 4 shipped
Mobile platforms Mobile platforms

FLUTTER FIXES PLATFORM VIEW CRASHES ON IOS, SWIFT TACKLES REGION ISOLATION DIAGNOSTICS

Flutter shipped critical fixes for backgrounding platform views and test flakiness on iOS, while Swift's compiler now surfaces isolation history for data race detection.

cbracken engine-flutter-autoroll gottesmm Anthony Latsis
mobile 19 shipped
Want every project, not just this one?