chore: remove openai dependency from providers (llamastack#3398)

# What does this PR do?

The openai package is already a dependency of the llama-stack project
itself, so let's the project dictate which openai version we need and
avoid potential breakage with unsatisfiable dependency resolution.

Signed-off-by: Sébastien Han <[email protected]>

Author: leseb

llama_stack/providers/registry/batches.py

@@ -13,7 +13,7 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.batches,
             provider_type="inline::reference",
-            pip_packages=["openai"],
+            pip_packages=[],
             module="llama_stack.providers.inline.batches.reference",
             config_class="llama_stack.providers.inline.batches.reference.config.ReferenceBatchesImplConfig",
             api_dependencies=[

llama_stack/providers/registry/inference.py

@@ -75,7 +75,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="vllm",
-                pip_packages=["openai"],
+                pip_packages=[],
                 module="llama_stack.providers.remote.inference.vllm",
                 config_class="llama_stack.providers.remote.inference.vllm.VLLMInferenceAdapterConfig",
                 description="Remote vLLM inference provider for connecting to vLLM servers.",
@@ -151,9 +151,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="databricks",
-                pip_packages=[
-                    "openai",
-                ],
+                pip_packages=[],
                 module="llama_stack.providers.remote.inference.databricks",
                 config_class="llama_stack.providers.remote.inference.databricks.DatabricksImplConfig",
                 description="Databricks inference provider for running models on Databricks' unified analytics platform.",
@@ -163,9 +161,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="nvidia",
-                pip_packages=[
-                    "openai",
-                ],
+                pip_packages=[],
                 module="llama_stack.providers.remote.inference.nvidia",
                 config_class="llama_stack.providers.remote.inference.nvidia.NVIDIAConfig",
                 description="NVIDIA inference provider for accessing NVIDIA NIM models and AI services.",
@@ -175,7 +171,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="runpod",
-                pip_packages=["openai"],
+                pip_packages=[],
                 module="llama_stack.providers.remote.inference.runpod",
                 config_class="llama_stack.providers.remote.inference.runpod.RunpodImplConfig",
                 description="RunPod inference provider for running models on RunPod's cloud GPU platform.",
@@ -207,7 +203,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="gemini",
-                pip_packages=["litellm", "openai"],
+                pip_packages=["litellm"],
                 module="llama_stack.providers.remote.inference.gemini",
                 config_class="llama_stack.providers.remote.inference.gemini.GeminiConfig",
                 provider_data_validator="llama_stack.providers.remote.inference.gemini.config.GeminiProviderDataValidator",
@@ -218,7 +214,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="vertexai",
-                pip_packages=["litellm", "google-cloud-aiplatform", "openai"],
+                pip_packages=["litellm", "google-cloud-aiplatform"],
                 module="llama_stack.providers.remote.inference.vertexai",
                 config_class="llama_stack.providers.remote.inference.vertexai.VertexAIConfig",
                 provider_data_validator="llama_stack.providers.remote.inference.vertexai.config.VertexAIProviderDataValidator",
@@ -248,7 +244,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="groq",
-                pip_packages=["litellm", "openai"],
+                pip_packages=["litellm"],
                 module="llama_stack.providers.remote.inference.groq",
                 config_class="llama_stack.providers.remote.inference.groq.GroqConfig",
                 provider_data_validator="llama_stack.providers.remote.inference.groq.config.GroqProviderDataValidator",
@@ -270,7 +266,7 @@ def available_providers() -> list[ProviderSpec]:
             api=Api.inference,
             adapter=AdapterSpec(
                 adapter_type="sambanova",
-                pip_packages=["litellm", "openai"],
+                pip_packages=["litellm"],
                 module="llama_stack.providers.remote.inference.sambanova",
                 config_class="llama_stack.providers.remote.inference.sambanova.SambaNovaImplConfig",
                 provider_data_validator="llama_stack.providers.remote.inference.sambanova.config.SambaNovaProviderDataValidator",

llama_stack/providers/registry/scoring.py

@@ -38,7 +38,7 @@ def available_providers() -> list[ProviderSpec]:
         InlineProviderSpec(
             api=Api.scoring,
             provider_type="inline::braintrust",
-            pip_packages=["autoevals", "openai"],
+            pip_packages=["autoevals"],
             module="llama_stack.providers.inline.scoring.braintrust",
             config_class="llama_stack.providers.inline.scoring.braintrust.BraintrustScoringConfig",
             api_dependencies=[

pyproject.toml

@@ -32,7 +32,7 @@ dependencies = [
     "jinja2>=3.1.6",
     "jsonschema",
     "llama-stack-client>=0.2.21",
-    "openai>=1.99.6",
+    "openai>=1.100.0",                                # for expires_after support
     "prompt-toolkit",
     "python-dotenv",
     "python-jose[cryptography]",
@@ -80,7 +80,6 @@ dev = [
 unit = [
     "sqlite-vec",
     "ollama",
-    "openai",
     "aiosqlite",
     "aiohttp",
     "psycopg2-binary>=2.9.0",
@@ -105,7 +104,6 @@ unit = [
 # separately. If you are using "uv" to execute your tests, you can use the "--group" flag to specify extra
 # dependencies.
 test = [
-    "openai>=1.100.0",  # for expires_after support
     "aiosqlite",
     "aiohttp",
     "torch>=2.6.0",