Practical guides to local AI inference and hardware configuration. Planned benchmarks are published only after their method and evidence are reviewed.
Lessons from migrating a local agent host to Q8 and a 262,144-token context, including why model identifiers, response limits and latency must be tested separately.
LM Studio Profiles: Quantization, Context and the Time to a Useful Answer Read More »
Sentinel ran a verified Q8 Glimmer instance across two 32GiB B70s. Three neutral one-shot coding tests returned no final code, exposing the gap between load success and usefulness.
Glimmer Q8 on Two B70s: It Loaded, but Delivered No Code Read More »
A September 6 recovery traced an unresponsive Hermes backend to one costly approval regex, then repaired its scan behavior without weakening the policy.
An alive process, an unresponsive agent: the Hermes regex outage Read More »
A dated Markdown knowledge system separates facts from plans and history, generates bounded context, and verifies a staged handoff without pretending it trains a model.
Versioned agent context: Markdown facts, receipts and a pointer-last handoff Read More »
How we repaired a private lab dashboard by separating machine identity, service reachability, fresh telemetry and explicitly authorized controls.
Building Fleet Command: What Online Actually Means Read More »
Two 32GB GPUs give Romeo room to serve local agents and larger context profiles. Here is the configuration, what changed, and what the evidence actually proves.
Romeo: Building a Local AI Workstation Around Two R9700s Read More »
The Qwen3.6-27B model occupies a specific niche in the local LLM landscape: it is dense enough to outperform most 14B-class models on coding and reasoning benchmarks, yet small enough to fit entirely within a single consumer GPU at moderate quantization levels. Released by Alibaba’s Qwen team in April 2026, it is the first open-weight variant
Running Qwen3.6-27B Locally: A Complete Inference Guide Read More »
