Skip to main content

Blog

Articles, guides, and insights on local artificial intelligence

5 articles

Qwen3.8-2.4T-A95B: the math of running 2.4T open-weight parameters, 95B active
ModelsAdvanced

Qwen3.8-2.4T-A95B: the math of running 2.4T open-weight parameters, 95B active

The first Qwen-Max-class open-weight model (2.4T parameters, 95B active) outgrows any consumer node: the math on weights, multi-node hardware, 36-month amortization, and power costs, and when running it locally makes sense.

iaLocale Team
November 22, 2026
SynthAPI: synthetic environments and scaling laws for training tool-use agents
Use casesAdvanced

SynthAPI: synthetic environments and scaling laws for training tool-use agents

Controlled synthetic environments with failures on a dial to train local tool-use agents: SynthAPI's scaling laws and the math — ~€1.60 of energy vs ~€158 of API for 10,000 tasks.

iaLocale Team
November 11, 2026
NVIDIA's Cosmos3-H-Dreams: real-time surgical simulation, and local robotics
Use casesAdvanced

NVIDIA's Cosmos3-H-Dreams: real-time surgical simulation, and local robotics

A 15B world model that generates robotic surgery video in real time on a single GPU: what it changes for local robotics.

iaLocale Team
November 10, 2026
Mac Studio M5 Ultra: practical 512GB guide - models that fit and GGUF setup
GuidesAdvanced

Mac Studio M5 Ultra: practical 512GB guide - models that fit and GGUF setup

What actually fits in the 512GB Mac Studio M5 Ultra: the 4-bit rule for MoE (active vs total params), three GGUF setup paths (Ollama, LM Studio, llama.cpp), why the KV cache is the real bottleneck, and when the $9,999 pays back against APIs.

iaLocale Team
October 20, 2026
vLLM: serving models fast
GuidesAdvanced

vLLM: serving models fast

vLLM is the open-source engine that turns a local model into a concurrent server: PagedAttention, continuous batching, installation and the key concepts for choosing when to use it instead of Ollama.

iaLocale Team
September 20, 2026