From 7d3b4e5ea3bb893e726e2a89c9726dc7e79b893d Mon Sep 17 00:00:00 2001
From: Jon Saad-Falcon <41205309+jonsaadfalcon@users.noreply.github.com>
Date: Tue, 24 Mar 2026 18:04:25 -0700
Subject: [PATCH] fix: add MLX engine support to Qwen3.5 model catalog entries

Fixes recommend_model() returning empty string on Apple Silicon
when MLX is the recommended engine. Also adds gguf_file and
mlx_repo download metadata, and estimated_download_gb helper.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 src/openjarvis/core/config.py                |  5 +++
 src/openjarvis/intelligence/model_catalog.py | 16 ++++++--
 tests/core/test_recommend_model.py           | 40 ++++++++++++++++++++
 3 files changed, 57 insertions(+), 4 deletions(-)

diff --git a/src/openjarvis/core/config.py b/src/openjarvis/core/config.py
index 5ca8fa77..33160232 100644
--- a/src/openjarvis/core/config.py
+++ b/src/openjarvis/core/config.py
@@ -248,6 +248,11 @@ def recommend_model(hw: HardwareInfo, engine: str) -> str:
     return ""
 
 
+def estimated_download_gb(parameter_count_b: float) -> float:
+    """Estimate download size in GB for Q4_K_M quantized model."""
+    return parameter_count_b * 0.5 * 1.1
+
+
 # ---------------------------------------------------------------------------
 # Configuration hierarchy
 # ---------------------------------------------------------------------------
diff --git a/src/openjarvis/intelligence/model_catalog.py b/src/openjarvis/intelligence/model_catalog.py
index d77c1740..ee36976f 100644
--- a/src/openjarvis/intelligence/model_catalog.py
+++ b/src/openjarvis/intelligence/model_catalog.py
@@ -45,11 +45,13 @@ BUILTIN_MODELS: List[ModelSpec] = [
         parameter_count_b=3.0,
         active_parameter_count_b=0.6,
         context_length=131072,
-        supported_engines=("ollama", "vllm", "llamacpp", "sglang"),
+        supported_engines=("ollama", "vllm", "llamacpp", "sglang", "mlx"),
         provider="alibaba",
         metadata={
             "architecture": "moe",
             "hf_repo": "Qwen/Qwen3.5-3B",
+            "gguf_file": "qwen3.5-3b-q4_k_m.gguf",
+            "mlx_repo": "mlx-community/Qwen3.5-3B-4bit",
         },
     ),
     ModelSpec(
@@ -58,11 +60,13 @@ BUILTIN_MODELS: List[ModelSpec] = [
         parameter_count_b=8.0,
         active_parameter_count_b=1.0,
         context_length=131072,
-        supported_engines=("ollama", "vllm", "llamacpp", "sglang"),
+        supported_engines=("ollama", "vllm", "llamacpp", "sglang", "mlx"),
         provider="alibaba",
         metadata={
             "architecture": "moe",
             "hf_repo": "Qwen/Qwen3.5-8B",
+            "gguf_file": "qwen3.5-8b-q4_k_m.gguf",
+            "mlx_repo": "mlx-community/Qwen3.5-8B-4bit",
         },
     ),
     ModelSpec(
@@ -71,11 +75,13 @@ BUILTIN_MODELS: List[ModelSpec] = [
         parameter_count_b=14.0,
         active_parameter_count_b=2.0,
         context_length=131072,
-        supported_engines=("ollama", "vllm", "llamacpp", "sglang"),
+        supported_engines=("ollama", "vllm", "llamacpp", "sglang", "mlx"),
         provider="alibaba",
         metadata={
             "architecture": "moe",
             "hf_repo": "Qwen/Qwen3.5-14B",
+            "gguf_file": "qwen3.5-14b-q4_k_m.gguf",
+            "mlx_repo": "mlx-community/Qwen3.5-14B-4bit",
         },
     ),
     ModelSpec(
@@ -223,11 +229,13 @@ BUILTIN_MODELS: List[ModelSpec] = [
         active_parameter_count_b=0.5,
         context_length=262144,
         min_vram_gb=3.0,
-        supported_engines=("ollama", "vllm", "sglang", "llamacpp"),
+        supported_engines=("ollama", "vllm", "sglang", "llamacpp", "mlx"),
         provider="alibaba",
         metadata={
             "architecture": "moe",
             "hf_repo": "Qwen/Qwen3.5-4B",
+            "gguf_file": "qwen3.5-4b-q4_k_m.gguf",
+            "mlx_repo": "mlx-community/Qwen3.5-4B-4bit",
         },
     ),
     ModelSpec(
diff --git a/tests/core/test_recommend_model.py b/tests/core/test_recommend_model.py
index 8f03d430..61600b69 100644
--- a/tests/core/test_recommend_model.py
+++ b/tests/core/test_recommend_model.py
@@ -115,3 +115,43 @@ class TestRecommendModelEdgeCases:
         # With ollama, 397b is excluded (only vllm, sglang)
         result = recommend_model(hw, "ollama")
         assert result == "qwen3.5:122b"
+
+
+class TestRecommendModelMlx:
+    """Apple Silicon (MLX) model recommendation."""
+
+    def test_apple_silicon_8gb_mlx(self) -> None:
+        hw = HardwareInfo(
+            platform="darwin",
+            ram_gb=8.0,
+            gpu=GpuInfo(vendor="apple", name="Apple M1", vram_gb=8.0, count=1),
+        )
+        result = recommend_model(hw, "mlx")
+        assert result == "qwen3.5:8b"
+
+    def test_apple_silicon_16gb_mlx(self) -> None:
+        hw = HardwareInfo(
+            platform="darwin",
+            ram_gb=16.0,
+            gpu=GpuInfo(vendor="apple", name="Apple M2", vram_gb=16.0, count=1),
+        )
+        result = recommend_model(hw, "mlx")
+        assert result == "qwen3.5:14b"
+
+    def test_apple_silicon_32gb_mlx(self) -> None:
+        hw = HardwareInfo(
+            platform="darwin",
+            ram_gb=32.0,
+            gpu=GpuInfo(vendor="apple", name="Apple M2 Pro", vram_gb=32.0, count=1),
+        )
+        result = recommend_model(hw, "mlx")
+        assert result == "qwen3.5:14b"
+
+    def test_apple_silicon_64gb_mlx(self) -> None:
+        hw = HardwareInfo(
+            platform="darwin",
+            ram_gb=64.0,
+            gpu=GpuInfo(vendor="apple", name="Apple M2 Max", vram_gb=64.0, count=1),
+        )
+        result = recommend_model(hw, "mlx")
+        assert result == "qwen3.5:14b"