Who contributed to Local LLMs on August 20, 2026?

18 developers shipped this update, including hoyyeva, dhiltgen, ParthSareen, Kevin Hopper, Xuan-Son Nguyen, Sigbjørn Skjæret, Gabe Goodhart, and Max Krasnyansky, and 10 more.

What were the notable Local LLMs updates?

server: don't wedge chat and generate on a mid-stream parser error, app: add desktop onboarding flow, and [Frontend] Move api_server.py out openai folder.

Local LLMs

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

Pick a date

Topics: Local LLMs AI / ML Full archive →

The Wire · Showcase

OLLAMA FIXES DEADLOCK THAT HUNG CHAT AND GENERATE

By RepoJournal · Filed 07:02 UTC on August 20, 2026 · About Local LLMs

18 people shipped this

hoyyeva @hoyyeva 1 cited

dhiltgen @dhiltgen 1 cited

noooop @noooop 1 cited

hnyls2002 @hnyls2002 1 cited

ParthSareen @ParthSareen

Kevin Hopper @kh0pper

Xuan-Son Nguyen @ngxson

Sigbjørn Skjæret @CISC

Gabe Goodhart @gabe-l-hart

Max Krasnyansky @max-krasnyansky

ywang96 @ywang96

AndreasKaratzas @AndreasKaratzas

jcotant-inferact @jcotant-inferact

A mid-stream parser error in Ollama's server silently wedged chat and generate requests, leaking goroutines and hanging retries until a fix landed overnight.

Ollama's server had a nasty bug: a parser error mid-stream wrote to an unbuffered channel that the consumer had already stopped reading, so the completion never returned, the goroutine leaked, and retrying the same prompt hung with no log output [1]. The fix, now merged, unblocks retries on failures. In the same repo, the desktop app is getting its first onboarding flow, guiding new users through sign-in, local-only, or skip [2], and the Qwen3.8 renderer now merges system messages correctly, avoiding dropped or duplicated prompt content . Over in vLLM, the OpenAI-compatible server entrypoint is deprecated in favor of `vllm server`, a breaking change that will affect scripts and Kubernetes manifests [3]. Meanwhile, SGLang reverted its DeepEPv2 ElasticBuffer MoE backend [4], a sign the feature wasn't ready for prime time. The stack is moving fast, but one pattern stands out: both Ollama and SGLang shipped fixes for hangs and reverts, so prioritize stability over bleeding-edge features today.

Action items

→ Update Ollama to include the parser-error fix before your next production deploy ollama/ollama [immediate]
→ Migrate your OpenAI-compatible server launch command to `vllm server` vllm-project/vllm [plan]
→ Pin SGLang to a build without the DeepEPv2 revert if you depend on that backend sgl-project/sglang [monitor]

References

[1] server: don't wedge chat and generate on a mid-stream parser error ↗ ollama/ollama
[2] app: add desktop onboarding flow ↗ ollama/ollama
[3] [Frontend] Move api_server.py out openai folder ↗ vllm-project/vllm
[4] Revert "[Feature] Add DeepEPv2 (ElasticBuffer) MoE A2A backend" ↗ sgl-project/sglang

Quick answers

What shipped in Local LLMs on August 20, 2026?: A mid-stream parser error in Ollama's server silently wedged chat and generate requests, leaking goroutines and hanging retries until a fix landed overnight. In total, 131 commits, 131 pull requests, and 6 releases landed.
Who contributed to Local LLMs on August 20, 2026?: 18 developers shipped this update, including hoyyeva, dhiltgen, ParthSareen, Kevin Hopper, Xuan-Son Nguyen, Sigbjørn Skjæret, Gabe Goodhart, and Max Krasnyansky, and 10 more.
What were the notable Local LLMs updates?: server: don't wedge chat and generate on a mid-stream parser error, app: add desktop onboarding flow, and [Frontend] Move api_server.py out openai folder.

DeepSeek V4 fuses shared experts, as vLLM ships a security guard

The biggest local-LLM repos are moving fast: DeepSeek V4's MoE kernel gets a massive fusion, while vLLM patches a security hole.

+12

llms 274 shipped

Local LLMs Aug 18

LLAMA.CPP SHIPS BAILINGMOE3 SPEC-DECODING SUPPORT

llama.cpp lands full BailingMoE3 support with speculative decoding, giving developers a native path for a key emerging sparse MoE family.

+10

llms 234 shipped

Local LLMs Aug 17

LLAMA.CPP SHIPS MODEL OPTIMIZATIONS WHILE VLLM AND SGLANG CHASE MEMORY AND SCHEDULING GAINS

Across 87 commits and 5 releases, the local LLM stack is grinding through the unglamorous work that makes inference faster: killing dead code, fixing kernel launches, and removing synchronization bottlenecks.

llms 178 shipped

Local LLMs Aug 16

LLAMA.CPP OVERHAULS INFERENCE FLAGS, VLLM PATCHES DEEPSEEK SPARSE MLA

llama.cpp deprecates mmap flags in favor of unified load-mode, while vllm fixes critical issues blocking DeepSeek-V4-Flash across all decode paths.

+11

llms 136 shipped

Elsewhere on the wire

AI Agents Aug 21

CLINE V4.1.11 SHIPS IMAGE GENERATION AND AGENT-UI FIXES

Cline's latest release lets models generate images mid-task, while LlamaIndex quietly breaks frontend tool handlers.

ai-agents 78 shipped

@CachyOS Aug 21

CachyOS Scheduler Hotfixes and Kernel Upgrades Lead the Day

A hotfix for the scx_cake scheduler and the latest Linux 7.2 kernel are rolling out across CachyOS repos.

infra 66 shipped

Elixir & Phoenix Aug 21

Phoenix 1.8.12 patches channel message drop

Phoenix shipped a patch release that fixes a channel bug silently dropping messages without a join_ref, and LiveView is quietly rewriting the rules of navigation.

elixir 30 shipped

Mobile platforms Aug 21

SWIFT GATES RUNTIME VARS, KOTLIN DROPS GRADLE 8.13

Swift's runtime now ignores safety-disabling environment variables in privileged processes, and JetBrains is cutting Kotlin's Gradle support at the knees.