feat: allow user to register model alias explicitly, tests

# What does this PR do?

Context: https://github.com/llamastack/llama-stack/discussions/3483

This PR enables the registering `provider_model_id` as the model identifier without breaking backward compatibility.


## Test Plan
todo
# What does this PR do?


## Test Plan

llama_stack/distributions/dell/dell.py

@@ -79,15 +79,15 @@ def get_distribution_template() -> DistributionTemplate:
     )
 
     inference_model = ModelInput(
-        model_id="${env.INFERENCE_MODEL}",
+        provider_model_id="${env.INFERENCE_MODEL}",
         provider_id="tgi0",
     )
     safety_model = ModelInput(
-        model_id="${env.SAFETY_MODEL}",
+        provider_model_id="${env.SAFETY_MODEL}",
         provider_id="tgi1",
     )
     embedding_model = ModelInput(
-        model_id="all-MiniLM-L6-v2",
+        provider_model_id="all-MiniLM-L6-v2",
         provider_id="sentence-transformers",
         model_type=ModelType.embedding,
         metadata={

llama_stack/distributions/dell/run-with-safety.yaml

@@ -103,18 +103,21 @@ inference_store:
   db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/dell}/inference_store.db
 models:
 - metadata: {}
-  model_id: ${env.INFERENCE_MODEL}
+  provider_model_id: ${env.INFERENCE_MODEL}
   provider_id: tgi0
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata: {}
-  model_id: ${env.SAFETY_MODEL}
+  provider_model_id: ${env.SAFETY_MODEL}
   provider_id: tgi1
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 384
-  model_id: all-MiniLM-L6-v2
+  provider_model_id: all-MiniLM-L6-v2
   provider_id: sentence-transformers
   model_type: embedding
+  use_provider_model_id_as_id: false
 shields:
 - shield_id: ${env.SAFETY_MODEL}
 vector_dbs: []

llama_stack/distributions/dell/run.yaml

@@ -99,14 +99,16 @@ inference_store:
   db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/dell}/inference_store.db
 models:
 - metadata: {}
-  model_id: ${env.INFERENCE_MODEL}
+  provider_model_id: ${env.INFERENCE_MODEL}
   provider_id: tgi0
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 384
-  model_id: all-MiniLM-L6-v2
+  provider_model_id: all-MiniLM-L6-v2
   provider_id: sentence-transformers
   model_type: embedding
+  use_provider_model_id_as_id: false
 shields: []
 vector_dbs: []
 datasets: []

llama_stack/distributions/meta-reference-gpu/meta_reference.py

@@ -73,11 +73,11 @@ def get_distribution_template() -> DistributionTemplate:
     )
 
     inference_model = ModelInput(
-        model_id="${env.INFERENCE_MODEL}",
+        provider_model_id="${env.INFERENCE_MODEL}",
         provider_id="meta-reference-inference",
     )
     embedding_model = ModelInput(
-        model_id="all-MiniLM-L6-v2",
+        provider_model_id="all-MiniLM-L6-v2",
         provider_id="sentence-transformers",
         model_type=ModelType.embedding,
         metadata={
@@ -85,7 +85,7 @@ def get_distribution_template() -> DistributionTemplate:
         },
     )
     safety_model = ModelInput(
-        model_id="${env.SAFETY_MODEL}",
+        provider_model_id="${env.SAFETY_MODEL}",
         provider_id="meta-reference-safety",
     )
     default_tool_groups = [

llama_stack/distributions/meta-reference-gpu/run-with-safety.yaml

@@ -116,18 +116,21 @@ inference_store:
   db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/meta-reference-gpu}/inference_store.db
 models:
 - metadata: {}
-  model_id: ${env.INFERENCE_MODEL}
+  provider_model_id: ${env.INFERENCE_MODEL}
   provider_id: meta-reference-inference
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata: {}
-  model_id: ${env.SAFETY_MODEL}
+  provider_model_id: ${env.SAFETY_MODEL}
   provider_id: meta-reference-safety
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 384
-  model_id: all-MiniLM-L6-v2
+  provider_model_id: all-MiniLM-L6-v2
   provider_id: sentence-transformers
   model_type: embedding
+  use_provider_model_id_as_id: false
 shields:
 - shield_id: ${env.SAFETY_MODEL}
 vector_dbs: []

llama_stack/distributions/meta-reference-gpu/run.yaml

@@ -106,14 +106,16 @@ inference_store:
   db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/meta-reference-gpu}/inference_store.db
 models:
 - metadata: {}
-  model_id: ${env.INFERENCE_MODEL}
+  provider_model_id: ${env.INFERENCE_MODEL}
   provider_id: meta-reference-inference
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 384
-  model_id: all-MiniLM-L6-v2
+  provider_model_id: all-MiniLM-L6-v2
   provider_id: sentence-transformers
   model_type: embedding
+  use_provider_model_id_as_id: false
 shields: []
 vector_dbs: []
 datasets: []

llama_stack/distributions/nvidia/nvidia.py

@@ -53,11 +53,11 @@ def get_distribution_template() -> DistributionTemplate:
         config=NVIDIAEvalConfig.sample_run_config(),
     )
     inference_model = ModelInput(
-        model_id="${env.INFERENCE_MODEL}",
+        provider_model_id="${env.INFERENCE_MODEL}",
         provider_id="nvidia",
     )
     safety_model = ModelInput(
-        model_id="${env.SAFETY_MODEL}",
+        provider_model_id="${env.SAFETY_MODEL}",
         provider_id="nvidia",
     )
 

llama_stack/distributions/nvidia/run-with-safety.yaml

@@ -96,13 +96,15 @@ inference_store:
   db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/nvidia}/inference_store.db
 models:
 - metadata: {}
-  model_id: ${env.INFERENCE_MODEL}
+  provider_model_id: ${env.INFERENCE_MODEL}
   provider_id: nvidia
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata: {}
-  model_id: ${env.SAFETY_MODEL}
+  provider_model_id: ${env.SAFETY_MODEL}
   provider_id: nvidia
   model_type: llm
+  use_provider_model_id_as_id: false
 shields:
 - shield_id: ${env.SAFETY_MODEL}
   provider_id: nvidia

llama_stack/distributions/nvidia/run.yaml

@@ -85,88 +85,88 @@ inference_store:
   db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/nvidia}/inference_store.db
 models:
 - metadata: {}
-  model_id: meta/llama3-8b-instruct
-  provider_id: nvidia
   provider_model_id: meta/llama3-8b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama3-70b-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama3-70b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.1-8b-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.1-8b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.1-70b-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.1-70b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.1-405b-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.1-405b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.2-1b-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.2-1b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.2-3b-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.2-3b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.2-11b-vision-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.2-11b-vision-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.2-90b-vision-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.2-90b-vision-instruct
-  model_type: llm
-- metadata: {}
-  model_id: meta/llama-3.3-70b-instruct
   provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
   provider_model_id: meta/llama-3.3-70b-instruct
-  model_type: llm
-- metadata: {}
-  model_id: nvidia/vila
   provider_id: nvidia
-  provider_model_id: nvidia/vila
   model_type: llm
+  use_provider_model_id_as_id: false
+- metadata: {}
+  provider_model_id: nvidia/vila
+  provider_id: nvidia
+  model_type: llm
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 2048
     context_length: 8192
-  model_id: nvidia/llama-3.2-nv-embedqa-1b-v2
-  provider_id: nvidia
   provider_model_id: nvidia/llama-3.2-nv-embedqa-1b-v2
+  provider_id: nvidia
   model_type: embedding
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 1024
     context_length: 512
-  model_id: nvidia/nv-embedqa-e5-v5
-  provider_id: nvidia
   provider_model_id: nvidia/nv-embedqa-e5-v5
+  provider_id: nvidia
   model_type: embedding
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 4096
     context_length: 512
-  model_id: nvidia/nv-embedqa-mistral-7b-v2
-  provider_id: nvidia
   provider_model_id: nvidia/nv-embedqa-mistral-7b-v2
+  provider_id: nvidia
   model_type: embedding
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 1024
     context_length: 512
-  model_id: snowflake/arctic-embed-l
-  provider_id: nvidia
   provider_model_id: snowflake/arctic-embed-l
+  provider_id: nvidia
   model_type: embedding
+  use_provider_model_id_as_id: false
 shields: []
 vector_dbs: []
 datasets: []

llama_stack/distributions/open-benchmark/run.yaml

@@ -136,30 +136,32 @@ inference_store:
   db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/open-benchmark}/inference_store.db
 models:
 - metadata: {}
-  model_id: gpt-4o
-  provider_id: openai
   provider_model_id: gpt-4o
+  provider_id: openai
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata: {}
-  model_id: claude-3-5-sonnet-latest
-  provider_id: anthropic
   provider_model_id: claude-3-5-sonnet-latest
+  provider_id: anthropic
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata: {}
-  model_id: gemini/gemini-1.5-flash
-  provider_id: gemini
   provider_model_id: gemini/gemini-1.5-flash
+  provider_id: gemini
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata: {}
-  model_id: meta-llama/Llama-3.3-70B-Instruct
-  provider_id: groq
   provider_model_id: groq/llama-3.3-70b-versatile
+  provider_id: groq
   model_type: llm
+  model_id: meta-llama/Llama-3.3-70B-Instruct
+  use_provider_model_id_as_id: false
 - metadata: {}
-  model_id: meta-llama/Llama-3.1-405B-Instruct
-  provider_id: together
   provider_model_id: meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo
+  provider_id: together
   model_type: llm
+  model_id: meta-llama/Llama-3.1-405B-Instruct
+  use_provider_model_id_as_id: false
 shields:
 - shield_id: meta-llama/Llama-Guard-3-8B
 vector_dbs: []

llama_stack/distributions/postgres-demo/postgres_demo.py

@@ -75,7 +75,7 @@ def get_distribution_template() -> DistributionTemplate:
 
     default_models = [
         ModelInput(
-            model_id="${env.INFERENCE_MODEL}",
+            provider_model_id="${env.INFERENCE_MODEL}",
             provider_id="vllm-inference",
         )
     ]
@@ -85,7 +85,7 @@ def get_distribution_template() -> DistributionTemplate:
         config=SentenceTransformersInferenceConfig.sample_run_config(),
     )
     embedding_model = ModelInput(
-        model_id="all-MiniLM-L6-v2",
+        provider_model_id="all-MiniLM-L6-v2",
         provider_id=embedding_provider.provider_id,
         model_type=ModelType.embedding,
         metadata={

llama_stack/distributions/postgres-demo/run.yaml

@@ -88,14 +88,16 @@ inference_store:
   password: ${env.POSTGRES_PASSWORD:=llamastack}
 models:
 - metadata: {}
-  model_id: ${env.INFERENCE_MODEL}
+  provider_model_id: ${env.INFERENCE_MODEL}
   provider_id: vllm-inference
   model_type: llm
+  use_provider_model_id_as_id: false
 - metadata:
     embedding_dimension: 384
-  model_id: all-MiniLM-L6-v2
+  provider_model_id: all-MiniLM-L6-v2
   provider_id: sentence-transformers
   model_type: embedding
+  use_provider_model_id_as_id: false
 shields:
 - shield_id: meta-llama/Llama-Guard-3-8B
 vector_dbs: []

llama_stack/distributions/template.py

@@ -114,7 +114,7 @@ def get_model_registry(
                 identifier = f"{provider_id}/{model_id}" if ids_conflict and provider_id not in model_id else model_id
                 models.append(
                     ModelInput(
-                        model_id=identifier,
+                        model_id=identifier if identifier != entry.provider_model_id else None,
                         provider_model_id=entry.provider_model_id,
                         provider_id=provider_id,
                         model_type=entry.model_type,

llama_stack/distributions/watsonx/watsonx.py

@@ -73,7 +73,7 @@ def get_distribution_template(name: str = "watsonx") -> DistributionTemplate:
     ]
 
     embedding_model = ModelInput(
-        model_id="all-MiniLM-L6-v2",
+        provider_model_id="all-MiniLM-L6-v2",
         provider_id="sentence-transformers",
         model_type=ModelType.embedding,
         metadata={