opengradient 0.5.0a3__tar.gz → 0.5.1__tar.gz

{opengradient-0.5.0a3/src/opengradient.egg-info → opengradient-0.5.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: opengradient
-Version: 0.5.0a3
+Version: 0.5.1
 Summary: Python SDK for OpenGradient decentralized model management & inference services
 Author-email: OpenGradient <kyle@vannalabs.ai>
 License-Expression: MIT

{opengradient-0.5.0a3 → opengradient-0.5.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "opengradient"
-version = "0.5.0a3"
+version = "0.5.1"
 description = "Python SDK for OpenGradient decentralized model management & inference services"
 authors = [{name = "OpenGradient", email = "kyle@vannalabs.ai"}]
 readme = "README.md"

{opengradient-0.5.0a3 → opengradient-0.5.1}/src/opengradient/client.py

@@ -61,6 +61,9 @@ DEFAULT_RETRY_DELAY_SEC = 1
 
 PRECOMPILE_CONTRACT_ADDRESS = "0x00000000000000000000000000000000000000F4"
 
+X402_PROCESSING_HASH_HEADER = "x-processing-hash"
+X402_PLACEHOLDER_API_KEY = "0x1234567890abcdef1234567890abcdef1234567890abcdef1234567890abcdef"
+
 class Client:
     _inference_hub_contract_address: str
     _blockchain: Web3
@@ -456,7 +459,11 @@ class Client:
             OpenGradientError: If the inference fails.
         """
         # Check if this is a local model or external
-        if not local_model and not self._is_local_model(model_cid):
+        # TODO (Kyle): separate TEE and Vanilla completion requests
+        if inference_mode == LlmInferenceMode.TEE:
+            if model_cid not in TEE_LLM:
+                return OpenGradientError("That model CID is not supported yet for TEE inference")
+
             return self._external_llm_completion(
                 model=model_cid,
                 prompt=prompt,
@@ -467,11 +474,11 @@ class Client:
         
         # Original local model logic
         def execute_transaction():
-            if inference_mode != LlmInferenceMode.VANILLA and inference_mode != LlmInferenceMode.TEE:
+            if inference_mode != LlmInferenceMode.VANILLA:
                 raise OpenGradientError("Invalid inference mode %s: Inference mode must be VANILLA or TEE" % inference_mode)
 
-            if inference_mode == LlmInferenceMode.TEE and model_cid not in [llm.value for llm in TEE_LLM]:
-                raise OpenGradientError("That model CID is not supported yet supported for TEE inference")
+            if model_cid not in [llm.value for llm in LLM]:
+                raise OpenGradientError("That model CID is not yet supported for inference")
 
             contract = self._blockchain.eth.contract(address=self._inference_hub_contract_address, abi=self._inference_abi)
 
@@ -525,7 +532,7 @@ class Client:
         api_key = self._get_api_key_for_model(model)
     
         if api_key:
-            print("External LLM completion using API key")
+            logging.debug("External LLM completions using API key")
             url = f"{self._llm_server_url}/v1/completions"
             
             headers = {
@@ -573,8 +580,7 @@ class Client:
             ) as client:
                 headers = {
                     "Content-Type": "application/json",
-                    # "Authorization": "Bearer special-key"
-                    "Authorization": "Bearer 0x1234567890abcdef1234567890abcdef1234567890abcdef1234567890abcdef"
+                    "Authorization": f"Bearer {X402_PLACEHOLDER_API_KEY}"
                 }
                 
                 payload = {
@@ -595,15 +601,11 @@ class Client:
                     result = json.loads(content.decode())
                     payment_hash = ""
 
-                    print("Payment response headers: ", response.headers)
-                    print("Payment response content: ", result)
-
-                    if "X-Payment-Response" in response.headers:
-                        payment_response = decode_x_payment_response(response.headers["X-Payment-Response"])
-                        payment_hash = payment_response["transaction"]
+                    if X402_PROCESSING_HASH_HEADER in response.headers:
+                        payment_hash = response.headers[X402_PROCESSING_HASH_HEADER]
                     
                     return TextGenerationOutput(
-                        transaction_hash="external",  # No blockchain transaction for external
+                        transaction_hash="external",
                         completion_output=result.get("completion"),
                         payment_hash=payment_hash
                     )
@@ -629,7 +631,7 @@ class Client:
 
     def llm_chat(
         self,
-        model_cid: str,  # Changed from LLM to str
+        model_cid: str,
         messages: List[Dict],
         inference_mode: LlmInferenceMode = LlmInferenceMode.VANILLA,
         max_tokens: int = 100,
@@ -662,7 +664,11 @@ class Client:
             OpenGradientError: If the inference fails.
         """
         # Check if this is a local model or external
-        if not local_model and not self._is_local_model(model_cid):
+        # TODO (Kyle): separate TEE and Vanilla completion requests
+        if inference_mode == LlmInferenceMode.TEE:
+            if model_cid not in TEE_LLM:
+                return OpenGradientError("That model CID is not supported yet for TEE inference")
+
             return self._external_llm_chat(
                 model=model_cid,
                 messages=messages,
@@ -675,11 +681,11 @@ class Client:
         
         # Original local model logic
         def execute_transaction():
-            if inference_mode != LlmInferenceMode.VANILLA and inference_mode != LlmInferenceMode.TEE:
+            if inference_mode != LlmInferenceMode.VANILLA:
                 raise OpenGradientError("Invalid inference mode %s: Inference mode must be VANILLA or TEE" % inference_mode)
-
-            if inference_mode == LlmInferenceMode.TEE and model_cid not in TEE_LLM:
-                raise OpenGradientError("That model CID is not supported yet supported for TEE inference")
+            
+            if model_cid not in [llm.value for llm in LLM]:
+                raise OpenGradientError("That model CID is not yet supported for inference")
 
             contract = self._blockchain.eth.contract(address=self._inference_hub_contract_address, abi=self._inference_abi)
 
@@ -768,7 +774,7 @@ class Client:
         api_key = self._get_api_key_for_model(model)
     
         if api_key:
-            print("External LLM completion using API key")
+            logging.debug("External LLM completion using API key")
             url = f"{self._llm_server_url}/v1/chat/completions"
             
             headers = {
@@ -821,7 +827,7 @@ class Client:
             ) as client:
                 headers = {
                     "Content-Type": "application/json",
-                    "Authorization": "Bearer special-key"
+                    "Authorization": f"Bearer {X402_PLACEHOLDER_API_KEY}"
                 }
                 
                 payload = {
@@ -846,16 +852,17 @@ class Client:
                     result = json.loads(content.decode())
 
                     payment_hash = ""
-                    print("Payment response headers: ", response.headers)
-                    print("Payment response content: ", result)
-                    if "X-Payment-Response" in response.headers:
-                        payment_response = decode_x_payment_response(response.headers["X-Payment-Response"])
-                        payment_hash = payment_response["transaction"]
+                    if X402_PROCESSING_HASH_HEADER in response.headers:
+                        payment_hash = response.headers[X402_PROCESSING_HASH_HEADER]
+                    
+                    choices = result.get("choices")
+                    if not choices:
+                        raise OpenGradientError(f"Invalid response: 'choices' missing or empty in {result}")
                     
                     return TextGenerationOutput(
                         transaction_hash="external",
-                        finish_reason=result["choices"][0].get("finish_reason"),
-                        chat_output=result["choices"][0].get("message"),
+                        finish_reason=choices[0].get("finish_reason"),
+                        chat_output=choices[0].get("message"),
                         payment_hash=payment_hash
                     )
                     

{opengradient-0.5.0a3 → opengradient-0.5.1}/src/opengradient/types.py

@@ -193,18 +193,65 @@ class Abi:
 class LLM(str, Enum):
     """Enum for available LLM models"""
 
+    # Existing open-source OG hosted models
     META_LLAMA_3_8B_INSTRUCT = "meta-llama/Meta-Llama-3-8B-Instruct"
     LLAMA_3_2_3B_INSTRUCT = "meta-llama/Llama-3.2-3B-Instruct"
     QWEN_2_5_72B_INSTRUCT = "Qwen/Qwen2.5-72B-Instruct"
     META_LLAMA_3_1_70B_INSTRUCT = "meta-llama/Llama-3.1-70B-Instruct"
     DOBBY_UNHINGED_3_1_8B = "SentientAGI/Dobby-Mini-Unhinged-Llama-3.1-8B"
     DOBBY_LEASHED_3_1_8B = "SentientAGI/Dobby-Mini-Leashed-Llama-3.1-8B"
-
+    
+    # OpenAI models via TEE
+    GPT_4_1_2025_04_14 = "gpt-4.1-2025-04-14"
+    GPT_4O = "gpt-4o"
+    O4_MINI = "o4-mini"
+    
+    # Anthropic models via TEE
+    CLAUDE_3_7_SONNET = "claude-3.7-sonnet"
+    CLAUDE_3_5_HAIKU = "claude-3.5-haiku"
+    CLAUDE_4_0_SONNET = "claude-4.0-sonnet"
+    
+    # Google models via TEE
+    GEMINI_2_5_FLASH = "gemini-2.5-flash"
+    GEMINI_2_5_PRO = "gemini-2.5-pro"
+    GEMINI_2_0_FLASH = "gemini-2.0-flash"
+    
+    # xAI Grok models via TEE
+    GROK_3_MINI_BETA = "grok-3-mini-beta"
+    GROK_3_BETA = "grok-3-beta"
+    GROK_2_1212 = "grok-2-1212"
+    GROK_2_VISION_LATEST = "grok-2-vision-latest"
+    GROK_4_1_FAST = "grok-4.1-fast"
+    GROK_4_1_FAST_NON_REASONING = "grok-4-1-fast-non-reasoning"
 
 class TEE_LLM(str, Enum):
     """Enum for LLM models available for TEE execution"""
-
-    META_LLAMA_3_1_70B_INSTRUCT = "meta-llama/Llama-3.1-70B-Instruct"
+    
+    # Existing (Currently turned off)
+    # META_LLAMA_3_1_70B_INSTRUCT = "meta-llama/Llama-3.1-70B-Instruct"
+    
+    # OpenAI models via TEE
+    GPT_4_1_2025_04_14 = "gpt-4.1-2025-04-14"
+    GPT_4O = "gpt-4o"
+    O4_MINI = "o4-mini"
+    
+    # Anthropic models via TEE
+    CLAUDE_3_7_SONNET = "claude-3.7-sonnet"
+    CLAUDE_3_5_HAIKU = "claude-3.5-haiku"
+    CLAUDE_4_0_SONNET = "claude-4.0-sonnet"
+    
+    # Google models via TEE
+    GEMINI_2_5_FLASH = "gemini-2.5-flash"
+    GEMINI_2_5_PRO = "gemini-2.5-pro"
+    GEMINI_2_0_FLASH = "gemini-2.0-flash"
+    
+    # xAI Grok models via TEE
+    GROK_3_MINI_BETA = "grok-3-mini-beta"
+    GROK_3_BETA = "grok-3-beta"
+    GROK_2_1212 = "grok-2-1212"
+    GROK_2_VISION_LATEST = "grok-2-vision-latest"
+    GROK_4_1_FAST = "grok-4.1-fast"
+    GROK_4_1_FAST_NON_REASONING = "grok-4-1-fast-non-reasoning"
 
 
 @dataclass

{opengradient-0.5.0a3 → opengradient-0.5.1/src/opengradient.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: opengradient
-Version: 0.5.0a3
+Version: 0.5.1
 Summary: Python SDK for OpenGradient decentralized model management & inference services
 Author-email: OpenGradient <kyle@vannalabs.ai>
 License-Expression: MIT