Promote audited AXQuant v2 artifact at stable repository name

Preserve backward-compatible repository identity; v1 remains tagged legacy-pre-v2 and this exact revision is tagged v2 after verification.

README.md

@@ -15,6 +15,7 @@ tags:
 - qwen3
 - 8bit
 - 8-bit
+- v2
 - embedding
 - sentence-similarity
 ---
@@ -28,26 +29,30 @@ the BF16 source model. The language path is quantized under AXQuant protection f
 > artifact-integrity records, but it does not publish measured quality, long-context, kernel-speed,
 > or MTP-speed evidence. Do not interpret the AXQ product label as a benchmark claim.
 
+> **Stable-name v2.** `main` serves the audited v2 artifact for backward compatibility. The same revision is tagged `v2`; the replaced artifact remains recoverable at `legacy-pre-v2`.
+
+
 ## Model details
 
 | Property | Value |
 | --- | --- |
-| Base model | [Qwen/Qwen3-Embedding-0.6B](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B) |
-| Source revision | `unrecorded` |
+| Base model | [Qwen/Qwen3-Embedding-0.6B](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B/tree/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3) |
+| Source revision | `97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3` |
 | Product family | `qwen3` |
 | Source architecture | `Qwen3ForCausalLM` (dense); text path optimized |
 | Main-model parameters | 595.78M logical parameters |
-| Quantizer | AXQuant `1.0.1` |
+| Quantizer | AXQuant `1.2.0` |
 | Hub budget class | `8bit` |
-| AXQuant base precision class | `4bit` |
+| Artifact edition | `v2` |
+| AXQuant base precision class | `8bit` |
 | Planned storage-adjusted BPW | 7.9992 |
 | Measured main-model BPW | 8.0003 |
 | Measured total BPW | **8.0003** |
 | Safetensors weight size | 0.60 GB |
 | Approximate complete download | 0.61 GB |
 | Configured maximum context | 32,768 tokens; practical limits depend on unified memory |
-| Primary runtime | AX Engine, compatibility level A |
-| Compatible runtime | MLX-LM standard text inference, compatibility level B |
+| MLX-LM compatibility | Standard text inference, compatibility level B |
+| AX Engine native execution | Not established; no validated native manifest is included |
 | MTP present | `False` |
 | Vision sidecar present | `False` |
 
@@ -65,7 +70,7 @@ protected models.
 | Sibling | Intended trade-off |
 | --- | --- |
 | [4bit sibling](https://huggingface.co/AutomatosX/AX-Qwen3-Embedding-0.6B-MLX-AXQ-4bit) | Lower-storage AXQ budget; check its exact BPW |
-| [6bit sibling](https://huggingface.co/AutomatosX/AX-Qwen3-Embedding-0.6B-MLX-AXQ-6bit) | Higher average precision near a 6-BPW budget |
+| [8bit sibling](https://huggingface.co/AutomatosX/AX-Qwen3-Embedding-0.6B-MLX-AXQ-8bit) | Higher average precision near the 8-BPW budget |
 
 See the [AutomatosX MLX model catalog](https://huggingface.co/collections/AutomatosX/automatosx-mlx-model-catalog)
 for related MLX and OptiQ alternatives.
@@ -96,19 +101,13 @@ metadata and optional sidecars (`vision.safetensors`, `mtp.safetensors`); this c
 does not establish MTP acceleration or vision-language quality. The artifact records MLX
 `0.32.0` and MLX-LM `0.31.3` from conversion.
 
-## Serve with AX Engine
+## AX Engine status
 
-After installing [AX Engine](https://github.com/defai-digital/ax-engine), download the complete
-repository and serve the local directory:
-
-```bash
-ax-engine serve ./AX-Qwen3-Embedding-0.6B-MLX-AXQ-8bit --port 31418
-```
-
-AX Engine is the authority for the AXQ runtime contract.
-This development package does not claim runtime speedups until identical-checkpoint benchmarks are
-published. The artifact records AX Engine version `not recorded`. Native
-`model-manifest.json` status: not included.
+This package does **not** include a validated native `model-manifest.json`, so AX Engine execution
+is not established by this release. The AX Engine fields in `axquant_runtime.json` describe the
+intended compatibility contract, not observed runtime evidence. Use the MLX-LM path above for
+standard text/backbone inference. The artifact records AX Engine version
+`not recorded`, but version discovery alone is not a runtime check.
 
 ## Quantization layout
 
@@ -151,6 +150,8 @@ establish MTP acceleration or vision-language quality.
 - Architecture-prior allocation is not measured sensitivity. It must not be presented as measured
   model quality.
 - The configured context window can require substantially more memory as the KV cache grows.
+- AX Engine execution is not established because this package has no validated native manifest.
+
 - Upstream capabilities, limitations, biases, and responsible-use guidance still apply.
 
 ## Provenance and audit files
@@ -160,7 +161,7 @@ establish MTP acceleration or vision-language quality.
 - [`axquant_plan.json`](axquant_plan.json): per-tensor precision decisions and planning evidence.
 - [`axquant_quantizer_execution.json`](axquant_quantizer_execution.json): conversion coverage and
   fallback records.
-- [`axquant_runtime.json`](axquant_runtime.json): AX Engine and MLX-LM compatibility contract.
+- [`axquant_runtime.json`](axquant_runtime.json): declared AX Engine and MLX-LM compatibility metadata; runtime checks remain separate evidence.
 
 All published provenance uses repository-relative paths. Local source paths are stripped before
 publication. The checkpoint was converted from BF16 rather than re-quantized from an OptiQ
@@ -170,5 +171,5 @@ identical BPW or quality.
 ## License
 
 The checkpoint follows the upstream model license where applicable (often Apache License 2.0). See
-the [Qwen/Qwen3-Embedding-0.6B model card](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B) for license terms, model
+the [Qwen/Qwen3-Embedding-0.6B model card](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B/tree/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3) for license terms, model
 limitations, and responsible-use guidance.

axquant_manifest.json

@@ -1,28 +1,28 @@
 {
-  "axquant_version": "1.0.1",
+  "axquant_version": "1.2.0",
   "calibration": null,
-  "created_at": "2026-08-02T19:19:15.735866Z",
+  "created_at": "2026-08-05T07:23:44.193379Z",
   "effective_bpw": 7.999220539932934,
   "files": [
     {
       "path": "README.md",
-      "sha256": "cf24e4d54458f4dc5c7725e37e7935422d2cdf2a6bd4875c580a8eeb9905ea69",
-      "size_bytes": 7338
+      "sha256": "7065c607c4251d4c917a1d75f6d99d7bfe690845ad101a7e3625e540cbb30841",
+      "size_bytes": 7792
     },
     {
       "path": "axquant_plan.json",
-      "sha256": "3238c60c26eb4a47866e944e8b53cfa31147cdb8676576fef5200e9f54081ed5",
-      "size_bytes": 292855
+      "sha256": "980597edd6f4f791380c00801791cbe653879d2bdbee5f92c25628497f548218",
+      "size_bytes": 292907
     },
     {
       "path": "axquant_quantizer_execution.json",
-      "sha256": "33426dabf2285479a2804dd0d516dd4051e521aab827a107f54001d2331da8bb",
+      "sha256": "f3f55fec94708899fae63efeb2b9dcdd8e3b9ee00c1ca3c08e23d4c2277bf1b9",
       "size_bytes": 48081
     },
     {
       "path": "axquant_runtime.json",
-      "sha256": "046b0b974cd0bb5b6ac407dbf57a53bcae885515e890b3cb87b7ac377f062ffc",
-      "size_bytes": 1643
+      "sha256": "1f554537b6ee3c3d28ea699ba55ac722eb816e33893a4cfab06c1cee8a41614a",
+      "size_bytes": 1628
     },
     {
       "path": "chat_template.jinja",
@@ -83,7 +83,7 @@
   },
   "mtp_present": false,
   "mtp_weight_file_size_bytes": 0,
-  "plan_sha256": "91a2de0eb8f6b3126ce18f8f832e835683fc929ad81d2397b906edeff714a293",
+  "plan_sha256": "3a2ba166158417fdaf008e097ac304e683383153205485c1db0bba1e2bcbaeea",
   "profile": "general",
   "protected_weight_file_size_bytes": 0,
   "quantizer": "axquant",
@@ -99,7 +99,7 @@
       {
         "compatibility_level": "B",
         "manifest": "config.json",
-        "mtp_support": "runtime-dependent",
+        "mtp_support": "none",
         "name": "mlx-lm",
         "notes": [
           "Standard backbone inference is the compatibility target.",
@@ -110,7 +110,7 @@
         "support_level": "standard-inference"
       }
     ],
-    "created_at": "2026-08-02T19:19:15.444403Z",
+    "created_at": "2026-08-05T07:23:43.990411Z",
     "kv_cache": null,
     "memory_policy": {
       "kv_cache_precision": "runtime-default",
@@ -134,7 +134,7 @@
     "primary_runtime": {
       "compatibility_level": "A",
       "manifest": "model-manifest.json",
-      "mtp_support": "native",
+      "mtp_support": "none",
       "name": "ax-engine",
       "notes": [
         "Runtime claims require a passing AX Engine doctor and benchmark report."
@@ -148,7 +148,7 @@
   "schema_version": "axquant.artifact.v2",
   "software_versions": {
     "ax_engine": null,
-    "axquant": "1.0.1",
+    "axquant": "1.2.0",
     "mlx": "0.32.0",
     "mlx_lm": "0.31.3",
     "pydantic": "2.13.4",
@@ -160,9 +160,9 @@
     "format": "mlx",
     "local_path": null,
     "model_id": "Qwen/Qwen3-Embedding-0.6B",
-    "revision": null
+    "revision": "97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3"
   },
-  "target_class": "4bit",
+  "target_class": "8bit",
   "weight_distribution": {
     "6bit": {
       "fraction": 0.3379230096268045,

axquant_plan.json

@@ -1,5 +1,5 @@
 {
-  "analysis_sha256": "6aa99da301ace53c56ce3c7e1293e9ed730362c5c8cb1c2d0598c6a9f09adf44",
+  "analysis_sha256": "62f2865677b2792a65c24900787c7278c465a99e4691917aae0a5dd5b33a3689",
   "architecture_profile": {
     "adapter_id": "qwen3-dense-v1",
     "config_model_type": "qwen3",
@@ -9027,7 +9027,7 @@
     "minimum_mtp_speedup": 1.2,
     "minimum_quality_retention": 0.98
   },
-  "created_at": "2026-08-02T19:19:13.447019Z",
+  "created_at": "2026-08-05T07:23:42.326212Z",
   "effective_bpw": 7.999220539932934,
   "evidence_kind": "architecture_prior",
   "global_validation_required": true,
@@ -9052,6 +9052,7 @@
       "affine",
       "awq",
       "dwq",
+      "gptq",
       "bf16"
     ]
   },
@@ -9089,7 +9090,7 @@
   "schema_version": "axquant.plan.v1",
   "software_versions": {
     "ax_engine": null,
-    "axquant": "1.0.1",
+    "axquant": "1.2.0",
     "mlx": "0.32.0",
     "mlx_lm": "0.31.3",
     "pydantic": "2.13.4",
@@ -9101,11 +9102,11 @@
     "format": "mlx",
     "local_path": null,
     "model_id": "Qwen/Qwen3-Embedding-0.6B",
-    "revision": null
+    "revision": "97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3"
   },
   "status": "planned",
   "target_bpw": 8.0,
-  "target_class": "4bit",
+  "target_class": "8bit",
   "target_mode": "low-memory",
   "warnings": [
     "This report contains architecture priors, not calibration measurements.",

axquant_quantizer_execution.json

@@ -1,6 +1,6 @@
 {
-  "created_at": "2026-08-02T19:19:15.440196Z",
-  "plan_sha256": "91a2de0eb8f6b3126ce18f8f832e835683fc929ad81d2397b906edeff714a293",
+  "created_at": "2026-08-05T07:23:43.986784Z",
+  "plan_sha256": "3a2ba166158417fdaf008e097ac304e683383153205485c1db0bba1e2bcbaeea",
   "records": [
     {
       "bits": 8,

axquant_runtime.json

@@ -10,7 +10,7 @@
     {
       "compatibility_level": "B",
       "manifest": "config.json",
-      "mtp_support": "runtime-dependent",
+      "mtp_support": "none",
       "name": "mlx-lm",
       "notes": [
         "Standard backbone inference is the compatibility target.",
@@ -21,7 +21,7 @@
       "support_level": "standard-inference"
     }
   ],
-  "created_at": "2026-08-02T19:19:15.444403Z",
+  "created_at": "2026-08-05T07:23:43.990411Z",
   "kv_cache": null,
   "memory_policy": {
     "kv_cache_precision": "runtime-default",
@@ -45,7 +45,7 @@
   "primary_runtime": {
     "compatibility_level": "A",
     "manifest": "model-manifest.json",
-    "mtp_support": "native",
+    "mtp_support": "none",
     "name": "ax-engine",
     "notes": [
       "Runtime claims require a passing AX Engine doctor and benchmark report."