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. Today 6 aug
Topics: Local LLMs AI / ML Full archive →

The Wire · Showcase

LLAMA.CPP SHIPS FOUR BUILDS IN 24 HOURS AS VLLM AND SGLANG RACE ON SEQUENCE PARALLELISM

By RepoJournal · Filed · About Local LLMs

github-actions[bot] Cheng Wan Isotr0py WoosukKwon yudigege86 Zhangmj0621 JustinTong0323
10 people shipped this
github-actions[bot] @github-actions[bot] 4 cited Cheng Wan @ch-wan 2 cited Isotr0py @Isotr0py 1 cited WoosukKwon @WoosukKwon 1 cited yudigege86 @yudigege86 1 cited Zhangmj0621 @Zhangmj0621 1 cited JustinTong0323 @JustinTong0323 1 cited Xuan-Son Nguyen @ngxson yzong-rh @yzong-rh guan404ming @guan404ming

The local LLM stack is moving fast: llama.cpp pushed builds b10219-b10223 with reasoning persistence and security updates, while vLLM landed sequence parallelism for DeepSeek V4 and sglang unified its radix cache for multi-turn agentic work.

Llama.cpp released four builds across the briefing window [1] [2] [3] [4], anchored by CLI persistence of reasoning content in chat history [3], which fixes a critical gap: the old code collected reasoning from the stream but only stored assistant messages, so prior thoughts couldn't be re-injected on later turns. In parallel, BoringSSL got bumped to 0.20260730.0 [2], and CI errors were fixed [1]. Over at vLLM, the bigger architectural shift landed: sequence parallelism for DeepSeek V4 using collective ops from Kimi K3, unifying multi-path encoder cuda graphs [5] [6], and a ROCm MLA mask fix [7] to prevent causality leaks in speculative verify blocks. SGLang addressed two config infrastructure debt items [9] [11] and shipped session-reference-aware radix cache [8], which keeps agentic workloads from evicting prefixes still in use by active sessions. Kimi K3 support landed in SGLang with reasoning, tool-calls, and native OpenAI serving [10]. All three repos are consolidating patterns: llama.cpp on reasoning preservation, vLLM on parallelism unification, sglang on config consistency and agentic cache semantics.

One email a day. Unsubscribe in one click.

Action items

References

  1. [1] b10223 ↗ ggml-org/llama.cpp
  2. [2] b10221 ↗ ggml-org/llama.cpp
  3. [3] b10219 ↗ ggml-org/llama.cpp
  4. [4] b10218 ↗ ggml-org/llama.cpp
  5. [5] [1/N] Unify multiple-path encoder cuda graph support ↗ vllm-project/vllm
  6. [6] [DSV4] Implement Sequence Parallelism ↗ vllm-project/vllm
  7. [7] [ROCm][MLA] Mask the AITER MLA small-head verify flatten causally ↗ vllm-project/vllm
  8. [8] feat: Session-reference-aware Unified Radix Cache for agentic multi-turn workloads ↗ sgl-project/sglang
  9. [9] config: route parallel config-leaf reads through get_parallel() (#33170) sgl-project/sglang
  10. [10] [Kimi K3] Add reasoning, tool-call, and OpenAI serving support ↗ sgl-project/sglang
  11. [11] test: recover the config-namespace-migration deferrals (#33171) sgl-project/sglang

Quick answers

What shipped in Local LLMs on August 2, 2026?
The local LLM stack is moving fast: llama.cpp pushed builds b10219-b10223 with reasoning persistence and security updates, while vLLM landed sequence parallelism for DeepSeek V4 and sglang unified its radix cache for multi-turn agentic work. In total, 39 commits, 39 pull requests, and 4 releases landed.
Who contributed to Local LLMs on August 2, 2026?
10 developers shipped this update, including Xuan-Son Nguyen, github-actions[bot], Isotr0py, WoosukKwon, yzong-rh, guan404ming, yudigege86, and Cheng Wan, and 2 more.
What were the notable Local LLMs updates?
b10223, b10221, and b10219.

More from Local LLMs

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
Local LLMs Local LLMs

OLLAMA FIXES RACE CONDITIONS AS LLAMA.CPP SHIPS BREAKING TTS CHANGES

Ollama patched critical data races across its progress tracking system while llama.cpp introduces incompatible Qwen3-TTS support, and vLLM lands kernel optimizations that cut latency on specialized hardware.

dhiltgen github-actions[bot] Xuan-Son Nguyen Evan Huus +11
llms 246 shipped
Local LLMs Local LLMs

OLLAMA REWIRES OPENAI COMPATIBILITY, LLAMA.CPP HARDENS SAMPLING

Ollama reworked its entire streaming protocol to match OpenAI's wire format chunk-for-chunk, unblocking drop-in replacement deployments.

BruceMacD Konrad Moren Xuan-Son Nguyen Pascal +11
llms 230 shipped
Local LLMs Local LLMs

DEEPSEEK V4 INFERENCE GETS 50% SPEEDUP; LLAMA.CPP AND VLLM BOTH SHIPPING OPTIMIZED PATHS

DeepSeek V4 speculative decode just went from nice-to-have to standard: llama.cpp landed MTP and DSpark support with roughly 50% speedup [ref:3], vLLM hardened its KV cache coordinator for the same models [ref:7], and Metal got native hyper-connection ops [ref:5].

Talha Adnan aldehir am17an ggerganov +8
llms 148 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 HARDENS ANTHROPIC TOOL CALLING AND GIT CREDENTIALS AS OPENHANDS SHIPS 1.10.0

LangChain fixed a critical credential leak in evaluation runs while Anthropic tool schemas get proper fallback handling, as three major agent frameworks push releases across the ecosystem.

langchain-oss-model-profiles[bot] Mason Daugherty John Kennedy ccurme +7
ai-agents 54 shipped
CachyOS @CachyOS

WIKI TRANSLATION ROLLBACK CHAOS, MIRROR STACK HOLDING STEADY

CachyOS wiki maintainers are caught in a revert loop over German translations while the status infrastructure logs 55 commits without incident.

Jinra Aarrayy buster-biopsys
infra 57 shipped
Elixir & Phoenix Elixir & Phoenix

ELIXIR BUMPS ZIZMOR SECURITY SCANNER TO 1.28.0

The Elixir repo updated its CI security action overnight, pulling in the latest zizmor static analysis tooling.

dependabot
elixir 1 shipped
.NET @dotnet

DOTNET FOUNDATION PULLS iOS METRICS INTO FOCUS AS RUNTIME HARDENS HTTP PARSING

Performance team fixed iOS size reporting to measure only distributable packages, while runtime clamped down on RFC 9112 compliance for chunked encoding.

matouskozak dotnet-maestro Tomas Weinfurt Adeel Mujahid +4
dotnet 72 shipped
Want every project, not just this one?