Update README.md

README.md

@@ -7,239 +7,185 @@ sdk: static
 pinned: true
 ---
 
-# Matrix.Corp 🚀
+# Matrix.Corp
 
-Welcome to **Matrix.Corp** — an independent AI research organization pushing the boundaries of what intelligence can be. We build specialized models, novel architectures, and entirely new paradigms for real-world use cases.
+Independent AI research organization building specialized models, frontier agentic systems, and new intelligence paradigms.
 
----
-
-## Our Philosophy
-
-We believe intelligence should be **purpose-built, accessible, and honest**. Every project in our ecosystem is designed from the ground up for a specific domain, user, and hardware target. We focus on novel architectures, hardware-aware optimization, emotional intelligence, and — with Vexa — a completely new computational paradigm that goes beyond AI entirely.
+**HuggingFace:** [Matrix-Corp](https://huggingface.co/Matrix-Corp) · **Founded by:** [Zandy-Wandy](https://huggingface.co/Zandy-Wandy) · **GitHub:** [zapgaming](https://github.com/zapgaming)
 
 ---
 
-## Projects & Model Families
+## Status Legend
 
-### 🔷 Vexa — Crystalline Intelligence Substrate
-*NOT AN AI MODEL — A new computational paradigm*
-*Runs on any laptop · Fully open source · Lume language*
-**Status: 🔴 Spec complete — Build in progress**
+| Badge | Meaning |
+|---|---|
+| 🟢 Released | Weights available, ready to use |
+| 🟡 Preview | Architecture published, training planned |
+| 🔴 Planned | Design complete, not yet built |
+| 🩵 Long-Term | Vision defined, major research ahead |
+| 🟣 Closed | Proprietary weights |
+| ⬛ Deprecated | Cancelled or superseded |
 
-> **Vexa is not a neural network. It is a Crystalline Intelligence Substrate** — a living, self-updating lattice of Glyphs that crystallizes knowledge in 10 minutes on any laptop, learns from the live web in real time, and runs on Ollama. No gradient descent. No GPU cluster. No frozen knowledge.
+---
 
-**Core components:**
-- 🔷 **Glyph Lattice** — structured meaning objects, not weights. Every Glyph has identity, typed relations, confidence, source references, and a decay function
-- ⚡ **Crystallization** — replaces training. 5-phase pipeline, 10 min, any 8-core CPU, no GPU needed
-- 🌐 **Real-Time Learning** — 3 live threads: Web Crystallizer, Interaction Crystallizer, Decay Monitor. Never goes stale
-- 💻 **Lume** — declarative-relational programming language where meaning is a first-class citizen
-- 🔌 **Vexa Bridge** — Ollama / vLLM / HuggingFace compatible adapter
+## Models
 
-| Component | Repo | Status |
-|---|---|---|
-| Vexa-V1 (Max density, ~10B Glyphs) | Matrix-Corp/Vexa-V1 | 🔴 Planned |
-| Vexa-Micro-V1 (Laptop, ~10M Glyphs, 4GB RAM) | Matrix-Corp/Vexa-Micro-V1 | 🔴 Planned |
-| Lume Language Spec + Parser | Matrix-Corp/Lume-Language-Spec | 🔴 Planned |
-| Vexa Bridge (Ollama/vLLM/HF adapter) | Matrix-Corp/Vexa-Bridge | 🔴 Planned |
-| Vexa Crystallizer Engine | Matrix-Corp/Vexa-Crystallizer | 🔴 Planned |
+### 🌌 Zenith — Reasoning + Emotional Intelligence
+**Status:** 🟡 Preview · **Target:** Tenstorrent Blackhole p300a
 
-**Glyph Lattice density tiers:**
+Transformer models with a built-in EQ Engine — a dedicated emotional intelligence layer that sits alongside the reasoning stack. Ring Attention (32K), MoE (12 experts top-2), Ollama + vLLM compatible.
 
-| Tier | Glyphs | RAM | Equivalent | Use Case |
-|---|---|---|---|---|
-| Nano | ~1M | 2GB | ~1B LLM | IoT, edge devices |
-| Micro | ~10M | 4GB | ~7B LLM | Laptops, Raspberry Pi |
-| Core | ~100M | 8GB | ~13B LLM | Consumer GPU |
-| Dense | ~1B | 16GB | ~30B LLM | Workstation |
-| Max | ~10B | 40GB | ~70B LLM | A100 / p300a |
+| Model | Params | Base | Link |
+|---|---|---|---|
+| Zenith-7B-V1 | 7B | Qwen2.5-Coder-7B | [→](https://huggingface.co/Matrix-Corp/Zenith-7b-V1) |
+| Zenith-28B-V1 | 28B | Qwen3.5-27B (Opus 4.6 distilled) | [→](https://huggingface.co/Matrix-Corp/Zenith-28b-p300-V1) |
+| Zenith-32B-V1 | 32B | DeepSeek-R1-Distill-Qwen-32B | [→](https://huggingface.co/Matrix-Corp/Zenith-32b-V1-Tenstorrent-Blackhole-p300) |
+| Zenith-70B-V1 | 70B | DeepSeek-R1-Distill-Llama-70B | [→](https://huggingface.co/Matrix-Corp/Zenith-70b-V1-Tenstorrent-Blackhole-p300) |
 
-**Collection:** Matrix-Corp/vexa-v1 *(coming soon)*
+[View Zenith Collection →](https://huggingface.co/collections/Matrix-Corp/zenith-v1)
 
 ---
 
-### ⚡ Kairiq — Critical Moment Intelligence Module
-*Universal plug-in module · Built entirely in Lume · Slap-on adapters*
-**Status: 🖤 SCRAPED AND DELETED**
-
-> **Kairiq is not a model. It is a universal intelligence amplifier** — a plug-in module built entirely in `.lume` files that attaches to any Matrix.Corp model and elevates it to elite benchmark performance. A Kairiq-enhanced 32B competes with a vanilla 70B. Not because it knows more — because it deploys what it knows at exactly the right moment, the right depth, and the right priority.
-
-**The Three KQ Dimensions:**
-- ⏱ **Temporal Acuity (T)** — sensing rhythm, pacing, momentum. Knowing when a reasoning path is collapsing before it wastes compute
-- ⚡ **Tension Sensing (X)** — reading pressure gradients. Identifying load-bearing sub-problems where failure cascades everywhere
-- 👑 **Imperial Hierarchy (H)** — the rank and weight of every sub-problem. What overrides what. The actual question beneath the stated question
-
-**Three-layer pipeline:** Kairiq Gate → Kairiq Router → Base Model → Kairiq Verifier
-
-**Benchmark targets:**
-
-| Benchmark | KQ Dimension | Expected Gain |
-|---|---|---|
-| MMLU | H — ranks actual question instantly | +4–6% |
-| HumanEval | T — kills dead reasoning paths early | +6–9% |
-| MATH | X — slow-deep on load-bearing steps | +5–8% |
-| ARC | H — no reasoning overkill on simple problems | +3–5% |
-| HellaSwag | T — reads narrative momentum correctly | +4–7% |
+### 🔬 Vortex Scientific — Deep Science Reasoning
+**Status:** 🟡 Preview · **Target:** MacBook M2/M3 + Nvidia 4060
 
-**Written entirely in `.lume`. Python `adapter.py` bridge for any existing model. Zero changes to base model required.**
+Built from scratch — no base model. Custom 50K science tokenizer. Hybrid SSM+Attention architecture with four domain-specific modules: Equation/LaTeX, Numerical, Citation, and Molecular/Periodic Table.
 
-| Component | Repo | Status |
+| Model | Params | Link |
 |---|---|---|
-| Kairiq Module V1 | Matrix-Corp/Kairiq-Module-V1 | 🔴 Planned |
-| Lume Kairiq Spec | Matrix-Corp/Lume-Kairiq-Spec | 🔴 Planned |
-| Zenith-32B-Kairiq-V1 | Matrix-Corp/Zenith-32B-Kairiq-V1 | 🔴 Planned |
+| Vortex-7B-V1 | 7B | [→](https://huggingface.co/Matrix-Corp/Vortex-7b-V1) |
+| Vortex-13B-V1 | 13B | [→](https://huggingface.co/Matrix-Corp/Vortex-13b-V1) |
 
-**Collection:** Matrix-Corp/kairiq-v1 *(coming soon)*
+[View Vortex Collection →](https://huggingface.co/collections/Matrix-Corp/vortex-v1)
 
 ---
 
-### 🌌 Zenith — Reasoning & Emotional Intelligence
-*Optimized for Tenstorrent Blackhole p300a hardware*
+### 🌿 Touch Grass — Music AI
+**Status:** 🟡 Preview · **Target:** Any hardware
 
-High-performance reasoning models built for the Tenstorrent p300a accelerator (dual-chip, 32 RISC-V cores, 64GB GDDR6). Ring Attention (32K context), Mixture of Experts, and EQ Engine V1 for emotional intelligence.
+LoRA fine-tune on Qwen3.5 built for musicians. Tab & Chord Module, Music Theory Engine, Ear Training, EQ Adapter (4 emotional modes), Songwriting Module.
 
-| Model | Parameters | Base Model | Status |
+| Model | Params | Base | Link |
 |---|---|---|---|
-| [Zenith-7B-V1](https://huggingface.co/Matrix-Corp/Zenith-7b-V1) | 7B | Qwen2.5-Coder-7B | 🟡 Preview |
-| [Zenith-28B-V1](https://huggingface.co/Matrix-Corp/Zenith-28b-p300-V1) | 28B | Qwen3.5-27B (Claude Opus 4.6 distilled) | 🟡 Preview |
-| [Zenith-32B-V1](https://huggingface.co/Matrix-Corp/Zenith-32b-p300-V1) | 32B | DeepSeek-R1-Distill-Qwen-32B | 🟡 Preview |
-| [Zenith-70B-V1](https://huggingface.co/Matrix-Corp/Zenith-70b-p300-V1) | 70B | DeepSeek-R1-Distill-Llama-70B | 🟡 Preview |
+| TouchGrass-3B | 3B | Qwen3.5-3B-Instruct | [→](https://huggingface.co/Matrix-Corp/TouchGrass-3b) |
+| TouchGrass-7B | 7B | Qwen3.5-7B-Instruct | [→](https://huggingface.co/Matrix-Corp/TouchGrass-7b) |
 
-**Key features:** EQ Engine V1 · Ring Attention 32K · MoE 12 experts top-2 · TP=8/PP=4 · Ollama + vLLM compatible
-
-**Collection:** [Zenith V1](https://huggingface.co/collections/Matrix-Corp/zenith-v1)
+[View Touch Grass Collection →](https://huggingface.co/collections/Matrix-Corp/touch-grass)
 
 ---
 
-### 🔬 Vortex Scientific — Deep Science Reasoning
-*Optimized for Apple Silicon & Nvidia 4060 · Built from scratch — no base model*
-
-From-scratch models for scientific reasoning across Physics, Mathematics, Chemistry, Biology, Earth Science, Space Science, and Zoology. Hybrid SSM + attention architecture with a custom 50K science tokenizer and 4 specialized science modules.
+### 🌐 Matrix Lattice — Frontier Agentic MoE
+**Status:** 🟢 Released · 🟣 Closed Source · **Target:** 4–32× H100 / Tenstorrent p300a
 
-| Model | Parameters | Architecture | Status |
-|---|---|---|---|
-| [Vortex-7B-V1](https://huggingface.co/Matrix-Corp/Vortex-7b-V1) | 7B | Hybrid SSM + Attention (60% SSM) | 🟡 Preview |
-| [Vortex-13B-V1](https://huggingface.co/Matrix-Corp/Vortex-13b-V1) | 13B | Hybrid SSM + Attention (50% SSM) | 🟡 Preview |
+**Shipped.** Our largest and most capable system. Frontier-scale mixture-of-experts with 17 custom intelligence modules including: EQ Engine V2, Multi-Agent Coordination Layer (MACL), Hierarchical Context Compression Engine (HCCE), Causal Reasoning Graph, Long-Horizon Task Planner, Confidence Calibration Head, Safety Reasoning Module (SRM), and more. 1M token context across all tiers.
 
-**Key features:** No base model · Custom science tokenizer · Equation/LaTeX · Molecular/Periodic Table modules · Laptop-first
+| Model | Total Params | Active Params | Experts | Context | Link |
+|---|---|---|---|---|---|
+| Lattice-120B | 120B | ~22B | 64 top-4 | 1M | [→](https://huggingface.co/Matrix-Corp/Lattice-120B-V1) |
+| Lattice-430B | 430B | ~38B | 128 top-4 | 1M | [→](https://huggingface.co/Matrix-Corp/Lattice-430B-V1) |
+| Lattice-671B | 671B | ~47B | 256 top-4 | 1M | [→](https://huggingface.co/Matrix-Corp/Lattice-671B-V1) |
 
-**Collection:** [Vortex V1](https://huggingface.co/collections/Matrix-Corp/vortex-v1)
+[View Lattice Collection →](https://huggingface.co/collections/Matrix-Corp/lattice-v1)
 
 ---
 
-### 🌿 Touch Grass — Music AI Assistant
-*Ultra lightweight · Runs on anything*
+### 🩸 Matrix ECHO — Living Error Memory
+**Status:** 🔴 Build In Progress · 🟢 Open Source · **Language:** Rust
 
-Fine-tuned music assistant for learning instruments, music theory, songwriting, ear training, and music history. Warm, encouraging, beginner-friendly.
+**The model that remembers how it was wrong.**
 
-| Model | Parameters | Base Model | Status |
-|---|---|---|---|
-| [TouchGrass-3B](https://huggingface.co/Matrix-Corp/TouchGrass-3b) | 3B | Qwen3.5-3B-Instruct | 🟡 Preview |
-| [TouchGrass-7B](https://huggingface.co/Matrix-Corp/TouchGrass-7b) | 7B | Qwen3.5-7B-Instruct | 🟡 Preview |
+ECHO is a 27B coding-focused LLM built on `Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled`, running fully in Rust via HuggingFace `candle`. Every correction it receives crystallizes into a **Scar** — a typed, weighted memory object stored in a live petgraph lattice.
 
-**Key features:** All instruments · Tab & chord generation · Music theory engine · Ear training · Songwriting assistant · Music EQ adapter
+Before every response, ECHO scans its Scar lattice for similar past mistakes. The more it's corrected, the harder it is to fool. Mistakes are not erased — they become assets.
 
-**Collection:** [Touch Grass](https://huggingface.co/collections/Matrix-Corp/touch-grass)
+**Core loop:**
+```
+prompt → pre-scan Scar lattice → inject caution context → generate → correction → new Scar forms
+```
 
----
+**Scar types:** Factual · Logical · Contextual · Hallucination · Overconfidence
 
-### 🌐 Matrix Lattice — Frontier Agentic MoE
-*Inference provider deployment · Closed source · Long-term roadmap*
-**Status: 🩵 Long-Term Roadmap · Spec complete**
+**Domain Weakness Map** — ECHO tracks which topics it's systematically weak in and suppresses confidence automatically in high-risk domains.
 
-Flagship frontier agentic + multimodal MoE family with 17 custom modules including EQ Engine V2, Multi-Agent Coordination Layer (MACL), Hierarchical Context Compression Engine (HCCE), and Safety Reasoning Module. Designed for inference provider deployment (Novita, Hyperbolic, Together, Fireworks).
+**OpenAI-compatible API** — drop-in via `POST /v1/chat/completions`. Corrections via `POST /v1/echo/correct`.
 
-| Model | Total Params | Active Params | Context | Status |
-|---|---|---|---|---|
-| Lattice-120B | 120B | ~22B | 1M tokens | 🩵 Long-Term Roadmap |
-| Lattice-430B | 430B | ~38B | 1M tokens | 🩵 Long-Term Roadmap |
-| Lattice-671B | 671B | ~47B | 1M tokens | 🩵 Long-Term Roadmap |
+| Model | Params | Base | Language |
+|---|---|---|---|
+| ECHO-27B-V1 | 27B | Qwen3.5-27B (Opus 4.6 distilled) | Rust + candle |
 
-**Status:** 🩵 Long-Term Roadmap · 🟣 Closed Source
+[View ECHO Collection →](https://huggingface.co/collections/Matrix-Corp/echo-v1)
 
 ---
 
 ### 🎨 Matrix Voxel — 3D Generation
-*Flow Matching · Triplane Latent · A100 40GB*
-**Status: 🔴 Planned — Architecture complete**
+**Status:** 🔴 Planned · **Target:** A100 40GB
 
-3D generation model family sharing a flow-matching DiT backbone (~2.3B) with task-specific decoder heads.
+Flow-matching DiT backbone (~2.3B) with task-specific decoder heads. Generates 3D meshes, environments, printable models, and NeRF/Gaussian Splatting outputs.
 
-| Model | Task | Outputs | Status |
+| Model | Task | Outputs | License |
 |---|---|---|---|
-| Voxel Atlas | World/environment generation | .vox, .obj, .usd | 🔴 Planned · 🟢 Open |
-| Voxel Forge | 3D mesh & asset generation | .obj, .glb, .fbx, .usdz | 🔴 Planned · 🟢 Open |
-| Voxel Cast | 3D printable generation | .stl, .step, .3mf | 🔴 Planned · 🟢 Open |
-| Voxel Lens | NeRF / Gaussian Splatting | .ply (3DGS), NeRF weights | 🔴 Planned · 🟢 Open |
-| Voxel Prime | All-in-one unified | All formats | 🔴 Planned · 🟣 Closed |
+| Voxel Atlas | World/environment gen | .vox, .obj, .usd | 🟢 Open |
+| Voxel Forge | 3D mesh & assets | .obj, .glb, .fbx, .usdz | 🟢 Open |
+| Voxel Cast | 3D printable | .stl, .step, .3mf | 🟢 Open |
+| Voxel Lens | NeRF / Gaussian Splatting | .ply (3DGS) | 🟢 Open |
+| Voxel Prime | Unified all-in-one | All formats | 🟣 Closed |
 
 ---
 
-### ⚒️ Matrix Axiom — Extreme Reasoning Code Intelligence
-*Reserved · Design in progress*
-**Status: 🔴 Reserved — Design starting**
+### 🔷 Matrix Vexa — Crystalline Intelligence Substrate
+**Status:** 🔴 Paused · 🟢 Open Source
 
-> Axiom is Matrix.Corp's planned extreme reasoning coding model. Hybrid architecture on a proven base with structurally enforced pre-code reasoning, a self-verify + rewrite loop, internal multi-agent trident (Architect · Artisan · Auditor), and a custom semantic code tokenizer. Targets #1 on HumanEval. Any hardware. Kairiq-enhanced at launch.
+Vexa is not a model. It is a new intelligence paradigm — a living lattice of **Glyphs** (structured meaning objects) that grows through **Crystallization** instead of training. 10 minutes on any CPU. No GPU required. Knowledge never goes stale — three background threads continuously update from the web, interactions, and decay.
 
-*Full spec coming soon.*
+Full paradigm definition and build prompt complete. Build paused, will resume.
 
----
-
-## Status Legend
-
-| Status | Meaning |
-|---|---|
-| 🟢 Released | Trained weights available, benchmarks published |
-| 🟡 Preview | Architecture published, training in progress or planned |
-| 🔴 Planned | Design complete, build not yet started |
-| 🩵 Long-Term Roadmap | Vision defined, significant research and compute required |
-| 🟣 Closed Source | Weights and training are proprietary |
+[View Vexa Collection →](https://huggingface.co/collections/Matrix-Corp/vexa-v1)
 
 ---
 
-## Hardware Philosophy
-
-We build for **accessible, affordable hardware** — not just cloud GPUs:
+### ⬛ ~~Kairiq — Critical Moment Intelligence Module~~
+**Status:** ⬛ Deprecated
 
-| Series | Target Hardware |
-|---|---|
-| Vexa | Any 8-core laptop — CPU only |
-| Kairiq | Any — universal plug-in |
-| Zenith | Tenstorrent Blackhole p300a |
-| Vortex | MacBook M2/M3 + Nvidia 4060 laptop |
-| Touch Grass | Any hardware |
-| Axiom | Any / multi-target |
-| Matrix Lattice | 16–32× H100 / p300a |
-| Matrix Voxel | A100 40GB |
+A Lume-native intelligence amplifier module designed to wrap Matrix models. Deprecated — the custom Lume language runtime exceeded practical build complexity. Core ideas (pre-scan, confidence suppression, domain routing) absorbed into ECHO.
 
 ---
 
-## Reserved Project Names
+## Paradigms
 
-- **Vexa** → Crystalline Intelligence Substrate
-- **Kairiq** → Critical Moment Intelligence Module
-- **Axiom** → Extreme Reasoning Code Intelligence
+| Name | Type | Status |
+|---|---|---|
+| Crystalline Intelligence (Vexa) | Non-neural knowledge substrate | 🔴 Paused |
+| Living Error Memory (ECHO) | Scar-based mistake crystallization | 🔴 Build In Progress |
+| Ferric Attention | Ownership-typed attention mechanism | 🩵 Research concept |
 
 ---
 
-## Organization
+## Reserved Names
 
-**Matrix.Corp** is an independent AI research organization.
+These names are allocated to specific projects. Not available for other uses.
 
-- 👤 Founded by [Zandy-Wandy](https://huggingface.co/Zandy-Wandy)
-- 🐙 GitHub: [zapgaming](https://github.com/zapgaming)
-- 🤗 HuggingFace: [Matrix-Corp](https://huggingface.co/Matrix-Corp)
+| Name | Allocated To |
+|---|---|
+| Vexa | Crystalline Intelligence Substrate |
+| ECHO | Living Error Memory LLM |
+| Axiom | Future extreme reasoning model (planned) |
+| Lume | Declarative-relational language for Vexa |
 
 ---
 
-## Collections
+## Licensing
 
-- [Zenith V1](https://huggingface.co/collections/Matrix-Corp/zenith-v1)
-- [Vortex V1](https://huggingface.co/collections/Matrix-Corp/vortex-v1)
-- [Touch Grass](https://huggingface.co/collections/Matrix-Corp/touch-grass)
-- Matrix-Corp/vexa-v1 *(coming soon)*
-- Matrix-Corp/kairiq-v1 *(coming soon)*
-- Matrix-Corp/voxel-v1 *(coming soon)*
+| Model Family | License |
+|---|---|
+| Zenith | Apache 2.0 |
+| Vortex | Apache 2.0 |
+| Touch Grass | Apache 2.0 |
+| Matrix Lattice | Proprietary |
+| Matrix ECHO | Apache 2.0 |
+| Matrix Voxel (open tiers) | Apache 2.0 |
+| Matrix Voxel Prime | Proprietary |
+| Vexa | Apache 2.0 |
 
 ---
 
-*Building the future of intelligence — one paradigm at a time.* 🚀
+*Matrix.Corp — building intelligence that knows its own limits.*