Experiments on open-source models to find
what holds up and what breaks.

Gemma 4 E4B vs the Gemma Family: Enterprise Benchmark Showdown

› Gemma 4 E4B (4B params) scored 83.6% overall — beating the 3x larger Gemma 3 12B (82.3%) across 8 enterprise suites.

Structured JSON Output from Small LLMs

› A well-instructed 2B model jumped from 30% to 90% compliance — outperforming models 3–4x its size.

Context Position Bias in Small LLMs

› Each architecture fails differently. Gemma-2B has strong recency bias (p=0.023). Llama-3B is completely flat (p=1.0).

RAG Compliance Week 4: 100% Recall

› v2 accuracy dropped from 68% to 65% — but blocks 7 more benign queries to eliminate the final 4 missed attacks.

NeMo Guardrails vs Prompt Injections

› NeMo: 55% recall. Llama Guard: 58% recall. Enforcement Engine: 93% recall on prompt injections.

ALTK‑Evolve: On‑the‑Job Learning for AI Agents

The next phase of enterprise AI

OpenAI outlines the next phase of enterprise AI, as adoption accelerates across industries with Frontier, ChatGPT Enterprise, Codex, and company-wide AI agents.

Introducing the Child Safety Blueprint

Discover OpenAI’s Child Safety Blueprint—a roadmap for building AI responsibly with safeguards, age-appropriate design, and collaboration to protect and…

Who Governs the Machine? A Machine Identity Governance Taxonomy (MIGT) for AI Systems Operating Across Enterprise and Geopolitical Boundaries

In-Place Test-Time Training