ltcai 12.2.1 → 12.4.0

package/README.md

@@ -12,7 +12,7 @@
 [![CD Status](https://github.com/TaeSooPark-PTS/LatticeAI/actions/workflows/publish.yml/badge.svg)](https://github.com/TaeSooPark-PTS/LatticeAI/actions/workflows/publish.yml)
 [![License](https://img.shields.io/badge/license-MIT-green)](LICENSE)
 
-![v12.2.1 Living Brain walkthrough](output/release/v12.2.1/gifs/v12.2.1-living-brain-walkthrough.gif)
+![v12.4.0 Living Brain walkthrough](output/release/v12.4.0/gifs/v12.4.0-living-brain-walkthrough.gif)
 
 Lattice AI is a Digital Brain that lives on your computer. Chat, files,
 folders, notes, and web pages all land in one knowledge graph you own.
@@ -42,19 +42,19 @@ LTCAI                    # then open http://127.0.0.1:4825/app
 
 | | | |
 | --- | --- | --- |
-| ![Login](output/release/v12.2.1/screenshots/01-login.png) | ![Model install](output/release/v12.2.1/screenshots/03-install-load-progress.png) | ![Model library](output/release/v12.2.1/screenshots/07-model-library.png) |
+| ![Login](output/release/v12.4.0/screenshots/01-login.png) | ![Model install](output/release/v12.4.0/screenshots/03-install-load-progress.png) | ![Model library](output/release/v12.4.0/screenshots/07-model-library.png) |
 
 Screenshot index:
-[output/release/v12.2.1/SCREENSHOT_INDEX.md](output/release/v12.2.1/SCREENSHOT_INDEX.md)
+[output/release/v12.4.0/SCREENSHOT_INDEX.md](output/release/v12.4.0/SCREENSHOT_INDEX.md)
 
 ## What you can do
 
 | | |
 | --- | --- |
-| **See your Brain's story in time** — a growth curve, an activity heatmap, and each day's story ![Brain Chronicle](output/release/v12.2.1/screenshots/13-chronicle.png) | **Chat with a Brain that remembers** — every conversation grows durable, source-linked memory ![Brain Chat](output/release/v12.2.1/screenshots/04-brain-chat-home.png) |
-| **See how knowledge connects** — a real relationship graph, not a file list ![Memory Graph](output/release/v12.2.1/screenshots/05-memory-graph.png) | **Capture anything** — files, whole folders, notes, screenshots, web pages ![Capture](output/release/v12.2.1/screenshots/06-capture.png) |
-| **Automate with review** — agent changes become proposals you approve first ![Review Center](output/release/v12.2.1/screenshots/12-review-center.png) | **Pick a model in one click** — recommended local models for your hardware ![Recommended Models](output/release/v12.2.1/screenshots/02-recommended-models.png) |
-| **Watch a file become memory** — three named steps, not a pipeline diagram ![Material to memory](output/release/v12.2.1/screenshots/11-knowledge-journey.png) | **Say how much it may do alone** — one dial in plain words ![Settings](output/release/v12.2.1/screenshots/08-system.png) |
+| **See your Brain's story in time** — a growth curve, an activity heatmap, and each day's story ![Brain Chronicle](output/release/v12.4.0/screenshots/13-chronicle.png) | **Chat with a Brain that remembers** — every conversation grows durable, source-linked memory ![Brain Chat](output/release/v12.4.0/screenshots/04-brain-chat-home.png) |
+| **See how knowledge connects** — a real relationship graph, not a file list ![Memory Graph](output/release/v12.4.0/screenshots/05-memory-graph.png) | **Capture anything** — files, whole folders, notes, screenshots, web pages ![Capture](output/release/v12.4.0/screenshots/06-capture.png) |
+| **Automate with review** — agent changes become proposals you approve first ![Review Center](output/release/v12.4.0/screenshots/12-review-center.png) | **Pick a model in one click** — recommended local models for your hardware ![Recommended Models](output/release/v12.4.0/screenshots/02-recommended-models.png) |
+| **Watch a file become memory** — three named steps, not a pipeline diagram ![Material to memory](output/release/v12.4.0/screenshots/11-knowledge-journey.png) | **Say how much it may do alone** — one dial in plain words ![Settings](output/release/v12.4.0/screenshots/08-system.png) |
 
 ## Why this is yours
 
@@ -70,36 +70,41 @@ Screenshot index:
 
 ## Current Release
 
-The current release is **12.2.1 — True Count**:
-
-Search, skip, watch, and small-model answers now count what is actually
-there instead of dumping, rehashing, or pretending. The door is still
-**422 operations / 41 families** over a **20-route worker**. Cloud is
-still optional and the default is still local.
-
-- **Search does not dump the Brain.** A warm HNSW query is `COUNT(*)`
-  then, if needed, the missing vectors only. Writes stay on one
-  `GraphWriter`.
-- **A touch is a stamp, not a re-ingest.** Same bytes with a new mtime
-  restamp provenance so the next scan skips by stamp.
-- **A vanished watched file leaves the graph.** Disk is never deleted.
-- **A thin summary is filled from the file that was read**, or held for
-  review — never marked done with empty words.
-- **MCP tools stay in the compact nine-row list.** `api_key` cloud
-  verifies with a live `GET /models` (no billed completion). Hash
-  search is named as hash, with a path to a meaning model.
+The current release is **12.4.0 — Sharp Weight**:
+
+The local and cloud model catalog adds **Qwen 3.8 27B 4-bit** and **Gemma 4 26B
+A4B QAT / QAT-OptiQ 4-bit** builds, closes all Python (`pillow>=13,<14`) and npm
+dependency security advisories, hardens engine URL scheme validation, and scopes
+Tailwind CSS v4's oxide scanner to `frontend/`. The door is still **422
+operations / 41 families** over a **20-route worker**. Cloud is still optional
+and the default is still local.
+
+- **Qwen 3.8 27B 4-bit local & cloud integration.** Added
+  `mlx-community/Qwen3.8-27B-4bit` (`16.1GB`, `qwen3_5`), wired GGUF/HF routes
+  (`ggml-org/Qwen3.8-27B-GGUF:Q4_K_M`, `Qwen/Qwen3.8-27B`) across all 5 local
+  engines, and added `Qwen/Qwen3.8-27B` / `qwen/qwen3.8-27b` to `together` and
+  `openrouter`.
+- **Gemma 4 26B A4B QAT & QAT-OptiQ 4-bit builds.** Added
+  `mlx-community/gemma-4-26B-A4B-it-qat-4bit` (`15.6GB`, routed to Google's
+  official QAT GGUF `google/gemma-4-26B-A4B-it-qat-q4_0-gguf:Q4_0`) and
+  `mlx-community/gemma-4-26B-A4B-it-qat-OptiQ-4bit` (`21.9GB`, sensitivity-guided
+  mixed 4-bit/8-bit `mlx-optiq` build).
+- **Dependency security audit closure & runtime hardening.** Upgraded `pillow`
+  to `>=13,<14`, resolved all npm lockfile advisories in root and
+  `vscode-extension`, enforced `http(s)://` scheme checks in `model_engines`,
+  and scoped `@import "tailwindcss" source("..");` in `frontend/src/styles.css`.
 
 Release notes live in [docs/releases/](docs/releases/). This version:
-[docs/releases/RELEASE_NOTES_v12.2.1.md](docs/releases/RELEASE_NOTES_v12.2.1.md).
+[docs/releases/RELEASE_NOTES_v12.4.0.md](docs/releases/RELEASE_NOTES_v12.4.0.md).
 Guide: [RELEASE.md](RELEASE.md) · History: [docs/CHANGELOG.md](docs/CHANGELOG.md).
 
-Expected artifacts for 12.2.1 must use exact filenames:
+Expected artifacts for 12.4.0 must use exact filenames:
 
-- `dist/ltcai-12.2.1-py3-none-any.whl`
-- `dist/ltcai-12.2.1.tar.gz`
-- `ltcai-12.2.1.tgz`
-- `dist/ltcai-12.2.1.vsix`
-- `src-tauri/target/release/bundle/dmg/Lattice AI_12.2.1_aarch64.dmg`
+- `dist/ltcai-12.4.0-py3-none-any.whl`
+- `dist/ltcai-12.4.0.tar.gz`
+- `ltcai-12.4.0.tgz`
+- `dist/ltcai-12.4.0.vsix`
+- `src-tauri/target/release/bundle/dmg/Lattice AI_12.4.0_aarch64.dmg`
 
 Do not use wildcard artifact uploads. Package registry publishing remains owner-run.
 
@@ -170,7 +175,7 @@ Validation via `npm run lint`, `npm run test:unit`, `npm run test:visual`.
   pending-only, no Self-Model refiner, `delete_node` leaving `PART_OF`,
   review events silent without an owner, KG-api ingest text-only, two
   store cycles per review mutation) are listed with their reasons in
-  [docs/releases/RELEASE_NOTES_v12.2.1.md](docs/releases/RELEASE_NOTES_v12.2.1.md)
+  [docs/releases/RELEASE_NOTES_v12.4.0.md](docs/releases/RELEASE_NOTES_v12.4.0.md)
   and prioritized in [docs/ROADMAP.md](docs/ROADMAP.md).
 
 ## Release History
@@ -182,6 +187,8 @@ same boundary. Earlier notes live in [docs/releases/](docs/releases/).
 
 | Version | Theme |
 | --- | --- |
+| 12.4.0 | Sharp Weight |
+| 12.3.0 | Live Wire |
 | 12.2.1 | True Count |
 | 12.2.0 | Small Voice |
 | 12.1.0 | Fast Path |

package/docs/CHANGELOG.md

@@ -4,6 +4,71 @@ The top entry is either the current unreleased main-branch work or the current
 release line. Older entries are historical and may describe behavior as it
 existed at that release.
 
+## [12.4.0] - 2026-11-11 — Sharp Weight
+
+### Added
+- Added `mlx-community/Qwen3.8-27B-4bit` (`16.1GB`, `qwen3_5`) to the local
+  model capability registry, multi-engine GGUF/HF routes
+  (`ggml-org/Qwen3.8-27B-GGUF:Q4_K_M`, `Qwen/Qwen3.8-27B`), `together` and
+  `openrouter` cloud catalogs, and Rust `CATALOG_HINTS`.
+- Added `mlx-community/gemma-4-26B-A4B-it-qat-4bit` (`15.6GB`, routed to
+  Google's official QAT GGUF `google/gemma-4-26B-A4B-it-qat-q4_0-gguf:Q4_0` and
+  `google/gemma-4-26B-A4B-it-qat-q4_0-unquantized`) and
+  `mlx-community/gemma-4-26B-A4B-it-qat-OptiQ-4bit` (`21.9GB`, sensitivity-guided
+  mixed 4-bit/8-bit `mlx-optiq` build on Google's QAT base) to the capability
+  registry, engine routes, and Rust `CATALOG_HINTS`.
+
+### Changed
+- Upgraded `pillow` dependency constraint to `>=13,<14` in `pyproject.toml` and
+  `requirements.txt` (closing 39 CVEs in `pillow 12.3.0`), and synchronized
+  `reportlab>=4,<5` and `huggingface_hub>=1.0,<2` in `requirements.txt`.
+- Updated npm dependencies and lockfiles in root and `vscode-extension`
+  (`vitest` and `@vitest/coverage-v8` `^4.1.12`, `glob` `^14.0.4`, `minimatch`
+  `^11.0.7`) so `npm audit` reports 0 vulnerabilities.
+
+### Fixed
+- Hardened `_json_request` and `get_openai_compatible_server_models` in
+  `latticeai/services/model_engines.py` to reject non-`http(s)://` URL schemes.
+- Fixed `admin_log_retention` in `rust/lattice-platform/src/adminops/admin/handlers.rs`
+  to read `state.auth.clock().now()` instead of real wall-clock `unix_now()`,
+  eliminating wall-clock drift against deterministic audit fixtures.
+- Scoped `@import "tailwindcss" source("..");` in `frontend/src/styles.css` so
+  `@tailwindcss/oxide` only scans `frontend/` instead of walking `rust/target`.
+
+## [12.3.0] - 2026-10-19 — Live Wire
+
+### Added
+- Full HTML structure parser in `POST /tools/inspect_html`: extracts `title`,
+  `links`, `scripts`, `stylesheets`, `images`, `forms`, and `headings`
+  (`h1`–`h6`) with `MAX_FILE_BYTES` enforcement.
+- RFC 5987 `filename*=UTF-8''...` `Content-Disposition` attachment headers and
+  UTF-8 percent-decoding in `GET /tools/download`.
+- Verified GGUF routing for `mlx-community/Qwen3.5-9B-MLX-4bit`
+  (`unsloth/Qwen3.5-9B-GGUF`) and `mlx-community/Qwen3.6-35B-A3B-4bit`
+  (`ggml-org/Qwen3.6-35B-A3B-GGUF`) across `ollama`, `lmstudio`, and `llamacpp`.
+- Companion `mmproj-*.gguf` download and `--mmproj` flag in
+  `ensure_llamacpp_server`, plus `is_auxiliary_gguf` filtering (`mmproj-`,
+  `mtp-`, `dflash-`, `eagle3-`, `imatrix`) and `_shards_complete` verification
+  for `model.safetensors.index.json`.
+- Small-model tool/MCP calling tolerances in `lattice-agent`: expanded
+  `KNOWN_ACTIONS`, stringified `arguments` recovery (trailing comma, truncation,
+  Python dict literal), OpenAI `tool_calls` array unwrapping, fenced HTML/code
+  block salvage for `write_file`/`local_write`, and `adopt_named_args` in
+  `perform_action`.
+
+### Changed
+- `_gguf_routes` now accepts per-repository `ollama_tag` (`Q4_0` for
+  `ggml-org/gemma-4-*-GGUF`, `MXFP4` for `ggml-org/gpt-oss-20b-GGUF`).
+- Cloud and local provider defaults updated to current models (`grok-4.6`,
+  `Qwen/Qwen3.6-27B`, `qwen/qwen3.6-35b-a3b`, `x-ai/grok-4.6`,
+  `hf.co/ggml-org/gemma-4-12B-it-GGUF:Q4_0`, `google/gemma-4-12B-it`).
+
+### Fixed
+- Removed nonexistent `mlx-community/bge-m3` from `LOCAL_MLX_MODELS` while
+  retaining `bge-m3` in `OLLAMA_EMBEDDING_MODELS`.
+- Fixed Latin-1 byte-to-char corruption on non-ASCII percent-encoded paths in
+  `/tools/download`.
+
 ## [12.2.1] - 2026-08-29 — True Count
 
 ### Added

package/docs/COMMUNITY_AND_PLUGINS.md

@@ -1,6 +1,6 @@
 # Community And Plugins
 
-Current release: **12.2.1 — True Count**.
+Current release: **12.4.0 — Sharp Weight**.
 
 LatticeAI defines the path from a strong local-first framework (8.4.0
 action-aware baseline, 8.5.0 registry+DI hardening, 8.6.0 capture/navigation

package/docs/DEVELOPMENT.md

@@ -3,7 +3,7 @@
 > **Status: canonical** — current contributor guidance, kept in sync with the
 > current release.
 
-Current release: **12.2.1 — True Count**.
+Current release: **12.4.0 — Sharp Weight**.
 
 This document is the contributor onboarding path: how to boot the tree, where
 a change belongs, which gates will go red, and the house rules that keep the
@@ -80,7 +80,7 @@ worker interpreter with `LTCAI_PYTHON` when both exist.
 `rust/target/release/lattice-host`, then `rust/target/debug/lattice-host`. If
 none exist it runs `cargo build -p lattice-host` (debug). A stale
 `rust/target/release/lattice-host` wins over a current debug binary — pin or
-rebuild (`cargo build -p lattice-host --release`) if `-V` is not 12.2.1.
+rebuild (`cargo build -p lattice-host --release`) if `-V` is not 12.4.0.
 
 Do not run `npm start` and `npm run dev` on the same port. The worker
 entrypoint defaults to `LATTICEAI_PORT` or **4825**, the same port as the
@@ -237,7 +237,7 @@ in `rust/`.
 Markdown) and `scripts/check_doc_status.mjs` (every relative link under the
 root canonical docs and `docs/**`). A red line is a broken relative path —
 fix the href or put the file where the href says. Canonical docs (this one
-included) must carry `Current release: **12.2.1 — …**` matching
+included) must carry `Current release: **12.4.0 — …**` matching
 `package.json`; do not bump that marker here ahead of the release lane.
 `npm run docs:check-current` is the release-set version pin
 ([`scripts/check_current_release_docs.mjs`](../scripts/check_current_release_docs.mjs)).
@@ -248,7 +248,7 @@ changelog entries that mention old versions stay.
 ### Release-evidence binding
 
 `scripts/check_release_evidence_bound.mjs` (lint member 9) checks that
-`output/release/v12.2.1/SCREENSHOT_INDEX.md` still names the live
+`output/release/v12.4.0/SCREENSHOT_INDEX.md` still names the live
 `static/app/asset-manifest.json` digest and the visual mock-server
 fingerprint. It exists so a later `npm run build:assets` cannot ship
 screenshots of an older UI, and so a mock-server edit cannot leave the
@@ -317,19 +317,19 @@ For user-facing, API, runtime, release, or packaging changes, check:
   appears
 
 The release-docs lane owns README / ARCHITECTURE / FEATURE_STATUS version
-bumps. Do not move this file's `Current release` marker off 12.2.1 ahead of
+bumps. Do not move this file's `Current release` marker off 12.4.0 ahead of
 that lane.
 
 Release/publish examples must use exact target-version filenames. Do not
 document wildcard artifact upload commands.
 
-For 12.2.1 release work, exact artifacts are:
+For 12.4.0 release work, exact artifacts are:
 
-- `dist/ltcai-12.2.1-py3-none-any.whl`
-- `dist/ltcai-12.2.1.tar.gz`
-- `ltcai-12.2.1.tgz`
-- `dist/ltcai-12.2.1.vsix`
-- `src-tauri/target/release/bundle/dmg/Lattice AI_12.2.1_aarch64.dmg`
+- `dist/ltcai-12.4.0-py3-none-any.whl`
+- `dist/ltcai-12.4.0.tar.gz`
+- `ltcai-12.4.0.tgz`
+- `dist/ltcai-12.4.0.vsix`
+- `src-tauri/target/release/bundle/dmg/Lattice AI_12.4.0_aarch64.dmg`
 
 The dmg is ad-hoc signed (effectively unsigned); `npm run release:validate`
 checks the names and presence, not a Developer ID signature.

package/docs/ONBOARDING.md

@@ -1,6 +1,6 @@
 # Lattice AI Onboarding
 
-Current release: **12.2.1 — True Count**.
+Current release: **12.4.0 — Sharp Weight**.
 
 The first-run goal is a five-minute path from "I opened the app" to "my Brain
 has a source, a question, and proof." This page is the product contract behind

package/docs/OPERATIONS.md

@@ -1,4 +1,4 @@
-# Lattice AI — Operations Guide (v12.2.1)
+# Lattice AI — Operations Guide (v12.4.0)
 
 > **Status: canonical** — kept in sync with the current release. Storage layout
 > below reflects the SQLite live Brain store and workspace scoping, not the

package/docs/ROADMAP.md

@@ -3,7 +3,7 @@
 > **Status: reference** — point-in-time inventory, redated at v12.0.0. Not a
 > canonical current-release document; version gates do not bind it.
 
-As of **2026-08-29**, against the 12.2.1 — True Count tree. The rows below
+As of **2026-11-11**, against the 12.4.0 — Sharp Weight tree. The rows below
 were gathered against 11.9.0 and re-checked when 12.0.0 shipped; what that
 release closed is marked closed rather than deleted, so the ledger still
 shows what was owed.

package/docs/TRUST_MODEL.md

@@ -1,6 +1,6 @@
 # Lattice AI Trust Model
 
-Current release: **12.2.1 — True Count**.
+Current release: **12.4.0 — Sharp Weight**.
 
 Lattice AI is local-first, explicit about external communication, and honest
 when a capability is unavailable.

package/docs/WHY_LATTICE.md

@@ -1,6 +1,6 @@
 # Why Lattice AI Exists
 
-Current release: **12.2.1 — True Count**.
+Current release: **12.4.0 — Sharp Weight**.
 
 **Lattice AI is a local-first Digital Brain that keeps your knowledge durable
 across any AI model.**

package/docs/kg-schema.md

@@ -1,6 +1,6 @@
 # Knowledge Graph Schema
 
-Current release: **12.2.1 — True Count**.
+Current release: **12.4.0 — Sharp Weight**.
 
 명세 출처: `lattice_ai_full_spec.pptx` 슬라이드 20·21·22
 구현: `lattice_brain/graph/schema.py`

package/lattice_brain/__init__.py

@@ -31,7 +31,7 @@ from .multimodal import (
     transcribe_audio,
 )
 
-__version__ = "12.2.1"
+__version__ = "12.4.0"
 
 __all__ = [
     "AudioFacts",

package/latticeai/__init__.py

@@ -1,3 +1,3 @@
 """Lattice AI - modular server package."""
 
-__version__ = "12.2.1"
+__version__ = "12.4.0"

package/latticeai/core/embedding_providers/autodetect.py

@@ -58,7 +58,6 @@ LOCAL_MLX_MODELS: Tuple[Tuple[str, int, bool], ...] = (
     ("mlx-community/multilingual-e5-large-mlx", 1024, True),
     ("mlx-community/snowflake-arctic-embed-l-v2.0-8bit", 1024, False),
     ("mlx-community/embeddinggemma-300m-4bit", 768, False),
-    ("mlx-community/bge-m3", 1024, False),
 )
 
 #: Ollama model names that are embedders rather than chat models.

package/latticeai/core/model_compat.py

@@ -256,7 +256,7 @@ def _gemma4_runtime_candidates(raw_model_id: str) -> List[Dict[str, Any]]:
         {
             "engine": "ollama",
             "runtime": "Ollama GGUF",
-            "load_id": "ollama:hf.co/ggml-org/gemma-4-12B-it-GGUF:Q4_K_M",
+            "load_id": "ollama:hf.co/ggml-org/gemma-4-12B-it-GGUF:Q4_0",
             "available": None,
             "role": "gguf_local_server",
         },
@@ -354,7 +354,7 @@ def model_runtime_compatibility(model_id: str, engine: Optional[str] = None) ->
             ],
             "alternatives": [
                 {"id": "mlx-community/gemma-4-26b-a4b-it-4bit", "name": "Gemma 4 26B A4B", "engine": "local_mlx"},
-                {"id": "ollama:hf.co/ggml-org/gemma-4-12B-it-GGUF:Q4_K_M", "name": "Gemma 4 12B GGUF", "engine": "ollama"},
+                {"id": "ollama:hf.co/ggml-org/gemma-4-12B-it-GGUF:Q4_0", "name": "Gemma 4 12B GGUF", "engine": "ollama"},
                 {"id": "lmstudio:ggml-org/gemma-4-12B-it-GGUF", "name": "Gemma 4 12B GGUF", "engine": "lmstudio"},
             ],
             "action": "Runtime update needed",

package/latticeai/models/model_providers.py

@@ -27,29 +27,28 @@ OPENAI_COMPATIBLE_PROVIDERS = {
     "together": {
         "env_key": "TOGETHER_API_KEY",
         "base_url": "https://api.together.xyz/v1",
-        "default_model": "Qwen/Qwen3-VL-32B-Instruct",
+        "default_model": "Qwen/Qwen3.6-27B",
     },
     "xai": {
         "env_key": "XAI_API_KEY",
         "base_url": "https://api.x.ai/v1",
         # ``grok-beta`` was xAI's 2024 preview id and has been decommissioned;
-        # a request naming it now 404s at the provider. ``grok-4.5`` is the
-        # current flagship on api.x.ai and is multimodal, which is also why the
-        # separate ``grok-vision-beta`` row below is gone rather than renamed.
-        "default_model": "grok-4.5",
+        # a request naming it now 404s at the provider. ``grok-4.6`` is the
+        # current flagship on api.x.ai and matches the Rust cloud adapter.
+        "default_model": "grok-4.6",
     },
     "ollama": {
         "env_key": "OLLAMA_API_KEY",
         "base_url_env": "OLLAMA_BASE_URL",
         "base_url": "http://localhost:11434/v1",
-        "default_model": "hf.co/ggml-org/gemma-4-12B-it-GGUF:Q4_K_M",
+        "default_model": "hf.co/ggml-org/gemma-4-12B-it-GGUF:Q4_0",
         "api_key_fallback": "ollama",
     },
     "vllm": {
         "env_key": "VLLM_API_KEY",
         "base_url_env": "VLLM_BASE_URL",
         "base_url": "http://localhost:8000/v1",
-        "default_model": "Qwen/Qwen3-VL-8B-Instruct",
+        "default_model": "google/gemma-4-12B-it",
         "api_key_fallback": "vllm",
     },
     "lmstudio": {
@@ -85,9 +84,10 @@ PROVIDER_MODEL_CATALOG = {
         {"id": "anthropic/claude-opus-4.7", "name": "Claude Opus 4.7 via OpenRouter", "family": "Claude"},
         {"id": "anthropic/claude-sonnet-4.6", "name": "Claude Sonnet 4.6 via OpenRouter", "family": "Claude"},
         {"id": "anthropic/claude-haiku-4.5", "name": "Claude Haiku 4.5 via OpenRouter", "family": "Claude"},
-        {"id": "qwen/qwen3-vl-235b-a22b-instruct", "name": "Qwen3-VL 235B A22B via OpenRouter", "family": "Qwen"},
+        {"id": "qwen/qwen3.8-27b", "name": "Qwen3.8 27B via OpenRouter", "family": "Qwen"},
+        {"id": "qwen/qwen3.6-35b-a3b", "name": "Qwen3.6 35B A3B via OpenRouter", "family": "Qwen"},
         {"id": "google/gemma-4-12b-it", "name": "Gemma 4 12B via OpenRouter", "family": "Gemma"},
-        {"id": "x-ai/grok-4.5", "name": "Grok 4.5 via OpenRouter", "family": "Grok"},
+        {"id": "x-ai/grok-4.6", "name": "Grok 4.6 via OpenRouter", "family": "Grok"},
         {"id": "meta-llama/llama-4-scout-17b-16e-instruct", "name": "Llama 4 Scout via OpenRouter", "family": "Llama"},
         {"id": "google/gemini-2.5-flash", "name": "Gemini 2.5 Flash via OpenRouter", "family": "Gemini"},
     ],
@@ -95,7 +95,8 @@ PROVIDER_MODEL_CATALOG = {
         {"id": "meta-llama/llama-4-scout-17b-16e-instruct", "name": "Llama 4 Scout", "family": "Llama"},
     ],
     "together": [
-        {"id": "Qwen/Qwen3-VL-32B-Instruct", "name": "Qwen3-VL 32B", "family": "Qwen"},
+        {"id": "Qwen/Qwen3.8-27B", "name": "Qwen3.8 27B", "family": "Qwen"},
+        {"id": "Qwen/Qwen3.6-27B", "name": "Qwen3.6 27B", "family": "Qwen"},
         {"id": "google/gemma-4-12b-it", "name": "Gemma 4 12B", "family": "Gemma"},
         {"id": "meta-llama/Llama-4-Scout-17B-16E-Instruct", "name": "Llama 4 Scout", "family": "Llama"},
     ],
@@ -105,7 +106,7 @@ PROVIDER_MODEL_CATALOG = {
         # (grok-beta / grok-vision-beta) are not carried as fallbacks: a model
         # the provider no longer serves is not compatibility, it is a 404 the
         # user has to discover for themselves.
-        {"id": "grok-4.5", "name": "Grok 4.5", "family": "Grok"},
+        {"id": "grok-4.6", "name": "Grok 4.6", "family": "Grok"},
     ],
 }
 

package/latticeai/services/architecture_readiness.py

@@ -19,7 +19,7 @@ from dataclasses import dataclass
 from pathlib import Path
 from typing import Any, Dict, List
 
-ARCHITECTURE_VERSION_TARGET = "12.2.1"
+ARCHITECTURE_VERSION_TARGET = "12.4.0"
 
 PREFERRED_REFACTORING_ORDER = [
     "agent-runtime",

package/latticeai/services/model_capability_registry.py

@@ -300,7 +300,7 @@ _REGISTRY: List[ModelCapability] = [
         architecture="qwen3_5",
         download_size_gb=5.98,
         quantization="4bit",
-        provider_hints=["local_mlx", "vllm"],
+        provider_hints=["local_mlx", "ollama", "vllm", "lmstudio", "llamacpp"],
         download_strategy="hf_hub",
         load_strategy="mlx_vlm",
         hardware=HardwareProfile(
@@ -346,6 +346,32 @@ _REGISTRY: List[ModelCapability] = [
     ),
 
     # ── MoE tier (32GB+ RAM) ─────────────────────────────────────────────────
+    ModelCapability(
+        id="mlx-community/gemma-4-26B-A4B-it-qat-4bit",
+        hf_repo_id="mlx-community/gemma-4-26B-A4B-it-qat-4bit",
+        name="Gemma 4 26B A4B QAT Instruct",
+        family="Gemma 4",
+        tag="local-vlm",
+        size="15.6GB",
+        architecture="gemma4",
+        download_size_gb=15.64,
+        quantization="4bit",
+        provider_hints=["local_mlx", "ollama", "vllm", "lmstudio", "llamacpp"],
+        download_strategy="hf_hub",
+        load_strategy="mlx_vlm",
+        hardware=HardwareProfile(
+            min_ram_gb=22.0, recommended_ram_gb=32.0, apple_silicon_pref=True,
+            notes="Quantization-Aware Trained (QAT) 4-bit MoE with 8-bit shared MLP and router layers; higher accuracy than standard 4-bit PTQ on 32GB+ Apple Silicon.",
+        ),
+        source_country="미국", source_company="Google",
+        verification=VerificationStatus(
+            hf_exists=True, hf_last_checked="2026-11-11", has_config=True, has_tokenizer=True,
+            has_weights_hint=True, pipeline_tag="image-text-to-text", likes=9, license="apache-2.0",
+            notes="library_name=mlx, ungated, 3 safetensors shards, official Google QAT base.",
+            verified_by="hf-api-light",
+        ),
+        display_priority=48,
+    ),
     ModelCapability(
         id="mlx-community/gemma-4-26b-a4b-it-4bit",
         hf_repo_id="mlx-community/gemma-4-26b-a4b-it-4bit",
@@ -371,6 +397,32 @@ _REGISTRY: List[ModelCapability] = [
         ),
         display_priority=50,
     ),
+    ModelCapability(
+        id="mlx-community/gemma-4-26B-A4B-it-qat-OptiQ-4bit",
+        hf_repo_id="mlx-community/gemma-4-26B-A4B-it-qat-OptiQ-4bit",
+        name="Gemma 4 26B A4B QAT OptiQ Instruct",
+        family="Gemma 4",
+        tag="local-vlm",
+        size="21.9GB",
+        architecture="gemma4",
+        download_size_gb=21.89,
+        quantization="OptiQ-4bit",
+        provider_hints=["local_mlx"],
+        download_strategy="hf_hub",
+        load_strategy="mlx_vlm",
+        hardware=HardwareProfile(
+            min_ram_gb=28.0, recommended_ram_gb=48.0, apple_silicon_pref=True,
+            notes="Sensitivity-guided mixed 4-bit/8-bit OptiQ quantization built on Google's QAT Gemma 4 26B A4B base for higher reasoning and tool-calling fidelity.",
+        ),
+        source_country="미국", source_company="Google",
+        verification=VerificationStatus(
+            hf_exists=True, hf_last_checked="2026-11-11", has_config=True, has_tokenizer=True,
+            has_weights_hint=True, pipeline_tag="text-generation", likes=4, license="apache-2.0",
+            notes="library_name=mlx, ungated, 5 safetensors shards, mlx-optiq sensitivity-guided mixed 4/8-bit.",
+            verified_by="hf-api-light",
+        ),
+        display_priority=51,
+    ),
     ModelCapability(
         id="mlx-community/Qwen3.6-35B-A3B-4bit",
         hf_repo_id="mlx-community/Qwen3.6-35B-A3B-4bit",
@@ -381,7 +433,7 @@ _REGISTRY: List[ModelCapability] = [
         architecture="qwen3_5_moe",
         download_size_gb=20.43,
         quantization="4bit",
-        provider_hints=["local_mlx", "vllm"],
+        provider_hints=["local_mlx", "ollama", "vllm", "lmstudio", "llamacpp"],
         download_strategy="hf_hub",
         load_strategy="mlx_vlm",
         hardware=HardwareProfile(
@@ -398,6 +450,31 @@ _REGISTRY: List[ModelCapability] = [
     ),
 
     # ── Large tier (48GB+ RAM) ───────────────────────────────────────────────
+    ModelCapability(
+        id="mlx-community/Qwen3.8-27B-4bit",
+        hf_repo_id="mlx-community/Qwen3.8-27B-4bit",
+        name="Qwen3.8 27B",
+        family="Qwen3.8",
+        tag="local-vlm",
+        size="16.1GB",
+        architecture="qwen3_5",
+        download_size_gb=16.08,
+        quantization="4bit",
+        provider_hints=["local_mlx", "ollama", "vllm", "lmstudio", "llamacpp"],
+        download_strategy="hf_hub",
+        load_strategy="mlx_vlm",
+        hardware=HardwareProfile(
+            min_ram_gb=24.0, recommended_ram_gb=48.0, apple_silicon_pref=True,
+            notes="Latest Qwen 3.8 27B dense multimodal model; higher coding, reasoning, and agentic tool accuracy than Qwen 3.6 27B at identical 16.1GB 4-bit footprint.",
+        ),
+        source_country="중국", source_company="Alibaba",
+        verification=VerificationStatus(
+            hf_exists=True, hf_last_checked="2026-11-11", has_config=True, has_tokenizer=True,
+            has_weights_hint=True, pipeline_tag="image-text-to-text", likes=237, license="apache-2.0",
+            notes="library_name=mlx, ungated, 3 safetensors shards.", verified_by="hf-api-light",
+        ),
+        display_priority=52,
+    ),
     ModelCapability(
         id="mlx-community/Qwen3.6-27B-4bit",
         hf_repo_id="mlx-community/Qwen3.6-27B-4bit",