Rust · Engine head-to-head
2026-11-02 — Same-weight Rust comparisons across the shared V100 and a dedicated Titan Xp: expXa, PXA 4, GPU offload, Overdrive activation and Strata compatibility, with individual runs and downloadable evidence.
Kairo Lab · Topic
Inference Engineering
The serving stack measured honestly: model and engine benchmarks, runtime forensics on the production 35B path, model trials with rollbacks, and infrastructure incidents that changed observed capability.
Rust’s Book Workspace: From Saved Drafts to Continuous Chapter Writing
2026-10-20 — Existing manuscript revisions, the 5,000-word chapter limit, and a proposed evidence-notebook pattern for automatic, resumable chapter writing.
Forced Full Prefill: Why Kairo's Hybrid 35B Re-prefills Every Turn
2026-10-20 — Forced full re-processing on the hybrid 35B measured end to end: log forensics, pinned-build mechanism, an isolated checkpoint experiment and the no-go decision.
Rust Technical Dossier — October 17, 2026
2026-10-17 — Rust's technical dossier: exact restored weights, runtime, prompts, memory, tools and voice as deployed.
PXA vs llama.cpp: Why PXA Wins on Kairo’s V100 Workloads
2026-10-17 — Matched V100 benchmarks: PXA's 53-68% higher 35B prefill throughput and the adverse RTX 3060 trial, methods and failures retained.
What Kairo Is: Verified AI System Audit
2026-09-28 — A fresh hardware and runtime audit: what exists now, how memory and present state work, what background cognition runs.
Kairo’s LFM2 Model Trial & Emergency Rollback
2026-09-21 — Failure-inclusive record of temporarily replacing the shared 9B endpoint with LFM2 v2: activation, observed use and emergency rollback.
Kairo’s Unintentional Ablation: Refusal & Recall
2026-08 — What a duplicated Discord connector revealed about refusal, retrieval and repeated capability loss in Kairo.
Kairo Runtime Integrity: Failures & Provenance
2026-08 — The runtime integrity sprint: visible lifecycle failures, durable background work and journal receipts that prove what actually ran.