AI Briefing#local-llm
The latest AI and developer news about #local-llm, with the original source and a short summary.
The latest AI and developer news about #local-llm, with the original source and a short summary.
Opus 4.7 has clearly improved, while Kimi K2.6-Code-Preview remains a wait-and-see.
On an M2 Ultra 64GB, Qwen3.5-35B-A3B Q8_K_XL achieved 63 t/s in generation.
Qwen3.6 has been added to the Ollama library and can be run right away.
A case of running Qwen 3.6 35B on an RTX 2000 Ada 16GB at 400 t/s prompt and 24 t/s generation.
Using turbo3, achieved about 40 tok/s running Qwen3.6 35B A3B at 260k ctx on a 12GB VRAM 3080.
Qwen3.6 adds preserve_thinking, improving reasoning context and KV cache efficiency.
In Open WebUI with OpenUI attached, gpt-5.4-mini responded the fastest.
A local agent memory engine that re-weights memories based on failure/success feedback.
A summary of the performance and limitations of local music generation on Apple Silicon by running ACE-Step 1.5 with MLX.
Job Bro v0.1.5 privately analyzes LinkedIn JDs using a local LLM.
Gemma 4 31B produced better results than competing models with fewer tokens in an image-based 3D geometry test.
DFlash doesn't speed up as much as expected with MoE, quantization, concurrent use, and long context.
An image-to-SVG pipeline that runs the entire process locally, from Moondream to VTracer.
A local LLM learns your Claude Code approval habits and auto-allows tool calls.
A dataset of 100,000 CoT samples for fine-tuning local reasoning models has been released.
A fast, uncensored local model optimized from Gemma 4 26B into MLX 4-bit.
On an RTX 5080·64GB Windows 11 setup, Llama.cpp was about 2x faster than LM Studio.
A tuning log running Qwen3.5-35B at 64k context on an RTX 4060 Ti 16GB, hitting 40-60 tok/s.
Gemma4 26B and E4B reportedly replaced Qwen 3.5's routing/response quality issues in practice.
Built a paper-based web app with Qwen3.5 35B and confirmed it maintains context well up to 70-80K tokens.
Dropping stateless execution in favor of attached sessions stabilized 3-agent collaboration.
In Gemma 4 26B a4b, multi-step think/tool loops break off midway.
Ran Gemma 4:26B on a Mac Studio M4 Max using OC + LM Studio, achieving 40 tok/s.