$(cat <<EOF

Add Groq provider support with ultra-fast inference models.

Groq API key validated and working.
Models added:
- groq/llama-3.3-70b-versatile (fast, 70B reasoning)
- groq/llama-3.1-8b-instant (fastest, 8B general)
- groq/qwen3-32b (fast, reasoning capable)

Files modified:
- config/provider_catalog.py: Added GROQ_DEFAULT_BASE and groq provider descriptor
- providers/defaults.py: Re-export GROQ_DEFAULT_BASE
- config/settings.py: Added groq_api_key and provider_is_configured check
- providers/registry.py: Added _create_groq factory
- providers/groq/client.py: New provider client
- providers/groq/__init__.py: New provider init
- core/model_capabilities.py: Registered groq models
- api/routes.py: Added groq models to REQUESTED_PROVIDER_MODELS
- .env: Added GROQ_API_KEY

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
EOF
)

.env

@@ -18,6 +18,7 @@ NVIDIA_NIM_FALLBACK_MODELS="nvidia_nim/qwen/qwen3-coder-480b-a35b-instruct,nvidi
 #CEREBRAS_API_KEY="csk-2ewy2h26eeph4yex94kmjnfwwx35pdpyyxkv3j6wcj4cxc3t"
 CEREBRAS_API_KEY="csk-2ewy2h26eeph4yex94kmjnfwwx35pdpyyxkv3j6wcj4cxc3t"
 SILICON_API_KEY="sk-vkswknrlhztbogulqjizbxpkdipbafudnirbrhzosxjkvmri"
+GROQ_API_KEY="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"
 
 LOG_RAW_API_PAYLOADS="true"
 # Disable API key authentication (allow requests without auth)

api/routes.py

@@ -52,6 +52,10 @@ REQUESTED_PROVIDER_MODELS = [
     "silicon/Qwen/Qwen3.5-27B",
     "silicon/google/gemma-4-26B-A4B-it",
     "silicon/google/gemma-4-31B-it",
+    # Groq models (ultra fast inference)
+    "groq/llama-3.3-70b-versatile",
+    "groq/llama-3.1-8b-instant",
+    "groq/qwen3-32b",
 ]
 
 

config/provider_catalog.py

@@ -16,6 +16,7 @@ NVIDIA_NIM_DEFAULT_BASE = "https://integrate.api.nvidia.com/v1"
 ZEN_DEFAULT_BASE = "https://opencode.ai/zen"
 CEREBRAS_DEFAULT_BASE = "https://api.cerebras.ai/v1"
 SILICON_DEFAULT_BASE = "https://api.siliconflow.com/v1"
+GROQ_DEFAULT_BASE = "https://api.groq.com/openai/v1"
 
 
 @dataclass(frozen=True, slots=True)
@@ -73,10 +74,19 @@ PROVIDER_CATALOG: dict[str, ProviderDescriptor] = {
         default_base_url=SILICON_DEFAULT_BASE,
         capabilities=("chat", "streaming", "tools", "thinking"),
     ),
+    "groq": ProviderDescriptor(
+        provider_id="groq",
+        transport_type="openai_chat",
+        credential_env="GROQ_API_KEY",
+        credential_url="https://console.groq.com/keys",
+        credential_attr="groq_api_key",
+        default_base_url=GROQ_DEFAULT_BASE,
+        capabilities=("chat", "streaming", "tools", "thinking"),
+    ),
 }
 
 # Order matches docs; must match PROVIDER_CATALOG keys.
-SUPPORTED_PROVIDER_IDS: tuple[str, ...] = ("nvidia_nim", "zen", "cerebras", "silicon")
+SUPPORTED_PROVIDER_IDS: tuple[str, ...] = ("nvidia_nim", "zen", "cerebras", "silicon", "groq")
 
 if len(set(SUPPORTED_PROVIDER_IDS)) != len(SUPPORTED_PROVIDER_IDS):
     raise AssertionError("Duplicate provider ids in PROVIDER_CATALOG key order")

config/settings.py

@@ -149,6 +149,8 @@ class Settings(BaseSettings):
     cerebras_api_key: str = Field(default="", validation_alias="CEREBRAS_API_KEY")
     # ==================== Silicon Flow Config ====================
     silicon_api_key: str = Field(default="", validation_alias="SILICON_API_KEY")
+    # ==================== Groq Config ====================
+    groq_api_key: str = Field(default="", validation_alias="GROQ_API_KEY")
     zen_base_url: str = Field(
         default="https://opencode.ai/zen", validation_alias="ZEN_BASE_URL"
     )
@@ -543,6 +545,8 @@ class Settings(BaseSettings):
             return bool(self.cerebras_api_key.strip())
         if provider_id == "silicon":
             return bool(self.silicon_api_key.strip())
+        if provider_id == "groq":
+            return bool(self.groq_api_key.strip())
         # conservative default: assume not configured
         return False
 

core/model_capabilities.py

@@ -228,6 +228,39 @@ MODEL_CAPABILITIES: dict[str, ModelCapabilities] = {
         speed="fast",
         priority=76,
     ),
+    # Groq models
+    "groq/llama-3.3-70b-versatile": ModelCapabilities(
+        provider_id="groq",
+        model_id="llama-3.3-70b-versatile",
+        model_ref="groq/llama-3.3-70b-versatile",
+        coding=True,
+        reasoning=True,
+        general_text=True,
+        max_tokens=32768,
+        speed="fast",
+        priority=85,
+    ),
+    "groq/llama-3.1-8b-instant": ModelCapabilities(
+        provider_id="groq",
+        model_id="llama-3.1-8b-instant",
+        model_ref="groq/llama-3.1-8b-instant",
+        coding=True,
+        general_text=True,
+        max_tokens=131072,
+        speed="fast",
+        priority=90,
+    ),
+    "groq/qwen3-32b": ModelCapabilities(
+        provider_id="groq",
+        model_id="qwen3-32b",
+        model_ref="groq/qwen3-32b",
+        coding=True,
+        reasoning=True,
+        general_text=True,
+        max_tokens=40960,
+        speed="medium",
+        priority=88,
+    ),
 }
 
 

providers/defaults.py

@@ -2,6 +2,7 @@
 
 from config.provider_catalog import (
     CEREBRAS_DEFAULT_BASE,
+    GROQ_DEFAULT_BASE,
     NVIDIA_NIM_DEFAULT_BASE,
     SILICON_DEFAULT_BASE,
     ZEN_DEFAULT_BASE,
@@ -9,6 +10,7 @@ from config.provider_catalog import (
 
 __all__ = (
     "CEREBRAS_DEFAULT_BASE",
+    "GROQ_DEFAULT_BASE",
     "NVIDIA_NIM_DEFAULT_BASE",
     "SILICON_DEFAULT_BASE",
     "ZEN_DEFAULT_BASE",

providers/groq/__init__.py

@@ -0,0 +1,5 @@
+"""Groq provider module."""
+
+from providers.groq.client import GroqProvider
+
+__all__ = ["GroqProvider"]

providers/groq/client.py

@@ -0,0 +1,41 @@
+"""Groq provider using OpenAI-compatible API."""
+
+from typing import Any
+
+from config.settings import Settings
+from core.anthropic import ReasoningReplayMode, build_base_request_body
+from providers.base import ProviderConfig
+from providers.defaults import GROQ_DEFAULT_BASE
+from providers.openai_compat import OpenAIChatTransport
+
+
+class GroqProvider(OpenAIChatTransport):
+    """Groq provider using OpenAI-compatible /chat/completions."""
+
+    def __init__(self, config: ProviderConfig, *, settings: Settings):
+        base_url = (config.base_url or GROQ_DEFAULT_BASE).rstrip("/")
+        if not base_url.endswith("/v1"):
+            base_url = base_url + "/v1"
+        super().__init__(
+            config,
+            provider_name="Groq",
+            base_url=base_url,
+            api_key=config.api_key,
+        )
+        self._settings = settings
+
+    def _build_request_body(
+        self, request: Any, thinking_enabled: bool | None = None
+    ) -> dict:
+        thinking = self._is_thinking_enabled(request, thinking_enabled)
+        reasoning_replay = (
+            ReasoningReplayMode.REASONING_CONTENT
+            if thinking
+            else ReasoningReplayMode.DISABLED
+        )
+        body = build_base_request_body(request, reasoning_replay=reasoning_replay)
+        # Strip groq/ prefix so the API gets the bare model ID
+        model = body.get("model", "")
+        if model.startswith("groq/"):
+            body["model"] = model[len("groq/") :]
+        return body

providers/registry.py

@@ -56,11 +56,18 @@ def _create_silicon(config: ProviderConfig, settings: Settings) -> BaseProvider:
     return SiliconProvider(config, settings=settings)
 
 
+def _create_groq(config: ProviderConfig, settings: Settings) -> BaseProvider:
+    from providers.groq import GroqProvider
+
+    return GroqProvider(config, settings=settings)
+
+
 PROVIDER_FACTORIES: dict[str, ProviderFactory] = {
     "nvidia_nim": _create_nvidia_nim,
     "zen": _create_zen,
     "cerebras": _create_cerebras,
     "silicon": _create_silicon,
+    "groq": _create_groq,
 }
 
 if set(PROVIDER_DESCRIPTORS) != set(SUPPORTED_PROVIDER_IDS) or set(