Cognitive Twin Architecture V2 — Decoupled RLM + Qwen 3.5 Migration

The Cognitive Twin is Mo's personal AI delegate — a model that knows his projects, preferences, reasoning patterns, and history. V1 used Llama 3.2:3B locally with a tightly-coupled RAG+Graph+RLM stack. V2 decouples every layer, swaps the base model to Qwen 3.5, and creates a clean evaluation pipeline. | Config | Score | What it proves | |--------|-------|---------------| | A: Bare Qwen3-Next-80B | 29.5% | API model has zero personal knowledge | | B: + RAG | 87.2% | Semantic retrieval is the biggest lever | | C: + Graph | 89.7% | Graph traversal adds marginal lift | | D: + RLM | 93.6% | RLM decomposition helps multi-hop queries | ### Layer 0: Base Model **Current:** Qwen3-Next-80B-A3B (Together AI, serverless, $0) **Target:** Qwen3.5-35B-A3B (local on Mac4+Mac5 exo cluster OR API) **Model Options (tested/available):** | Model | Where | Active Params | Score (Config A) | Cost | |-------|-------|---------------|------------------|------| | Llama 3.2:3B | Mac4 Ollama | 3B | ~25% | $0 | | Qwen3-Next-80B-A3B | Together API | 3B | 29.5% | $0 | | Qwen3.5-35B-A3B | OpenRouter | 3B | TBD | $0.16/M | | Qwen3.5-35B-A3B | Mac4+Mac5 exo | 3B | TBD | $0 | | Qwen3.5-397B-A17B | Together API | 17B | TBD | $0.60/M | **Migration path:** Start with API for evaluation speed, migrate to local exo cluster for $0 inference at scale.