opengradient 0.5.7__py3-none-any.whl → 0.5.9__py3-none-any.whl

opengradient/defaults.py

@@ -9,4 +9,6 @@ DEFAULT_BLOCKCHAIN_EXPLORER = "https://explorer.opengradient.ai/tx/"
 DEFAULT_IMAGE_GEN_HOST = "18.217.25.69"
 DEFAULT_IMAGE_GEN_PORT = 5125
 DEFAULT_LLM_SERVER_URL = "http://35.225.197.84:8000"
-DEFAULT_OPENGRADIENT_LLM_SERVER_URL = "https://llm.opengradient.ai"
+DEFAULT_OPENGRADIENT_LLM_SERVER_URL = "https://llmogevm.opengradient.ai"
+DEFAULT_OPENGRADIENT_LLM_STREAMING_SERVER_URL = "https://llmogevm.opengradient.ai"
+DEFAULT_NETWORK_FILTER = "og-evm"

opengradient/llm/og_langchain.py

@@ -36,7 +36,12 @@ class OpenGradientChatModel(BaseChatModel):
         super().__init__()
 
         self._client = Client(
-            private_key=private_key, rpc_url=DEFAULT_RPC_URL, api_url=DEFAULT_API_URL, contract_address=DEFAULT_INFERENCE_CONTRACT_ADDRESS, email=None, password=None
+            private_key=private_key,
+            rpc_url=DEFAULT_RPC_URL,
+            api_url=DEFAULT_API_URL,
+            contract_address=DEFAULT_INFERENCE_CONTRACT_ADDRESS,
+            email=None,
+            password=None,
         )
         self._model_cid = model_cid
         self._max_tokens = max_tokens

opengradient/types.py

@@ -1,15 +1,42 @@
 import time
 from dataclasses import dataclass
 from enum import Enum, IntEnum, StrEnum
-from typing import Dict, List, Optional, Tuple, Union, DefaultDict
+from typing import Dict, List, Optional, Tuple, Union, DefaultDict, Iterator, AsyncIterator
 import numpy as np
 
 
 class x402SettlementMode(StrEnum):
+    """
+    Settlement modes for x402 payment protocol transactions.
+
+    These modes control how inference data is recorded on-chain for payment settlement
+    and auditability. Each mode offers different trade-offs between data completeness,
+    privacy, and transaction costs.
+
+    Attributes:
+        SETTLE: Individual settlement with input/output hashes only.
+            Records cryptographic hashes of the inference input and output.
+            Most privacy-preserving option - actual data is not stored on-chain.
+            Suitable for applications where only proof of execution is needed.
+
+        SETTLE_METADATA: Individual settlement with full metadata.
+            Records complete model information, full input and output data,
+            and all inference metadata on-chain.
+            Provides maximum transparency and auditability.
+            Higher gas costs due to larger data storage.
+
+        SETTLE_BATCH: Batch settlement for multiple inferences.
+            Aggregates multiple inference requests into a single settlement transaction
+            using batch hashes.
+            Most cost-efficient for high-volume applications.
+            Reduced per-inference transaction overhead.
+    """
+
     SETTLE = "settle"
     SETTLE_METADATA = "settle-metadata"
     SETTLE_BATCH = "settle-batch"
 
+
 class CandleOrder(IntEnum):
     ASCENDING = 0
     DESCENDING = 1
@@ -107,7 +134,7 @@ class InferenceMode(Enum):
 
 
 class LlmInferenceMode(Enum):
-    """Enum for differetn inference modes available for LLM inferences (VANILLA, TEE)"""
+    """Enum for different inference modes available for LLM inference (VANILLA, TEE)"""
 
     VANILLA = 0
     TEE = 1
@@ -138,6 +165,196 @@ class InferenceResult:
     model_output: Dict[str, np.ndarray]
 
 
+@dataclass
+class StreamDelta:
+    """
+    Represents a delta (incremental change) in a streaming response.
+    
+    Attributes:
+        content: Incremental text content (if any)
+        role: Message role (appears in first chunk)
+        tool_calls: Tool call information (if function calling is used)
+    """
+    content: Optional[str] = None
+    role: Optional[str] = None
+    tool_calls: Optional[List[Dict]] = None
+
+
+@dataclass
+class StreamChoice:
+    """
+    Represents a choice in a streaming response.
+    
+    Attributes:
+        delta: The incremental changes in this chunk
+        index: Choice index (usually 0)
+        finish_reason: Reason for completion (appears in final chunk)
+    """
+    delta: StreamDelta
+    index: int = 0
+    finish_reason: Optional[str] = None
+
+
+@dataclass
+class StreamUsage:
+    """
+    Token usage information for a streaming response.
+    
+    Attributes:
+        prompt_tokens: Number of tokens in the prompt
+        completion_tokens: Number of tokens in the completion
+        total_tokens: Total tokens used
+    """
+    prompt_tokens: int
+    completion_tokens: int
+    total_tokens: int
+
+
+@dataclass
+class StreamChunk:
+    """
+    Represents a single chunk in a streaming LLM response.
+    
+    This follows the OpenAI streaming format but is provider-agnostic.
+    Each chunk contains incremental data, with the final chunk including
+    usage information.
+    
+    Attributes:
+        choices: List of streaming choices (usually contains one choice)
+        model: Model identifier
+        usage: Token usage information (only in final chunk)
+        is_final: Whether this is the final chunk (before [DONE])
+    """
+    choices: List[StreamChoice]
+    model: str
+    usage: Optional[StreamUsage] = None
+    is_final: bool = False
+    
+    @classmethod
+    def from_sse_data(cls, data: Dict) -> "StreamChunk":
+        """
+        Parse a StreamChunk from SSE data dictionary.
+        
+        Args:
+            data: Dictionary parsed from SSE data line
+            
+        Returns:
+            StreamChunk instance
+        """
+        choices = []
+        for choice_data in data.get("choices", []):
+            delta_data = choice_data.get("delta", {})
+            delta = StreamDelta(
+                content=delta_data.get("content"),
+                role=delta_data.get("role"),
+                tool_calls=delta_data.get("tool_calls")
+            )
+            choice = StreamChoice(
+                delta=delta,
+                index=choice_data.get("index", 0),
+                finish_reason=choice_data.get("finish_reason")
+            )
+            choices.append(choice)
+        
+        usage = None
+        if "usage" in data:
+            usage_data = data["usage"]
+            usage = StreamUsage(
+                prompt_tokens=usage_data.get("prompt_tokens", 0),
+                completion_tokens=usage_data.get("completion_tokens", 0),
+                total_tokens=usage_data.get("total_tokens", 0)
+            )
+        
+        is_final = any(c.finish_reason is not None for c in choices) or usage is not None
+        
+        return cls(
+            choices=choices,
+            model=data.get("model", "unknown"),
+            usage=usage,
+            is_final=is_final
+        )
+
+
+@dataclass
+class TextGenerationStream:
+    """
+    Iterator wrapper for streaming text generation responses.
+    
+    Provides a clean interface for iterating over stream chunks with
+    automatic parsing of SSE format.
+    
+    Usage:
+        stream = client.llm_chat(..., stream=True)
+        for chunk in stream:
+            if chunk.choices[0].delta.content:
+                print(chunk.choices[0].delta.content, end="")
+    """
+    _iterator: Union[Iterator[str], AsyncIterator[str]]
+    _is_async: bool = False
+    
+    def __iter__(self):
+        """Iterate over stream chunks."""
+        return self
+    
+    def __next__(self) -> StreamChunk:
+        """Get next stream chunk."""
+        import json
+        
+        while True:
+            try:
+                line = next(self._iterator)
+            except StopIteration:
+                raise
+            
+            if not line or not line.strip():
+                continue
+            
+            if not line.startswith("data: "):
+                continue
+            
+            data_str = line[6:]  # Remove "data: " prefix
+            
+            if data_str.strip() == "[DONE]":
+                raise StopIteration
+            
+            try:
+                data = json.loads(data_str)
+                return StreamChunk.from_sse_data(data)
+            except json.JSONDecodeError:
+                # Skip malformed chunks
+                continue
+    
+    async def __anext__(self) -> StreamChunk:
+        """Get next stream chunk (async version)."""
+        import json
+        
+        if not self._is_async:
+            raise TypeError("Use __next__ for sync iterators")
+        
+        while True:
+            try:
+                line = await self._iterator.__anext__()
+            except StopAsyncIteration:
+                raise
+            
+            if not line or not line.strip():
+                continue
+            
+            if not line.startswith("data: "):
+                continue
+            
+            data_str = line[6:]
+            
+            if data_str.strip() == "[DONE]":
+                raise StopAsyncIteration
+            
+            try:
+                data = json.loads(data_str)
+                return StreamChunk.from_sse_data(data)
+            except json.JSONDecodeError:
+                continue
+
+
 @dataclass
 class TextGenerationOutput:
     """
@@ -205,23 +422,23 @@ class LLM(str, Enum):
     # 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 = "openai/gpt-4.1-2025-04-14"
     GPT_4O = "openai/gpt-4o"
     O4_MINI = "openai/o4-mini"
-    
+
     # Anthropic models via TEE
     CLAUDE_3_7_SONNET = "anthropic/claude-3.7-sonnet"
     CLAUDE_3_5_HAIKU = "anthropic/claude-3.5-haiku"
     CLAUDE_4_0_SONNET = "anthropic/claude-4.0-sonnet"
-    
+
     # Google models via TEE
     GEMINI_2_5_FLASH = "google/gemini-2.5-flash"
     GEMINI_2_5_PRO = "google/gemini-2.5-pro"
     GEMINI_2_0_FLASH = "google/gemini-2.0-flash"
     GEMINI_2_5_FLASH_LITE = "google/gemini-2.5-flash-lite"
-    
+
     # xAI Grok models via TEE
     GROK_3_MINI_BETA = "x-ai/grok-3-mini-beta"
     GROK_3_BETA = "x-ai/grok-3-beta"
@@ -230,28 +447,29 @@ class LLM(str, Enum):
     GROK_4_1_FAST = "x-ai/grok-4.1-fast"
     GROK_4_1_FAST_NON_REASONING = "x-ai/grok-4-1-fast-non-reasoning"
 
+
 class TEE_LLM(str, Enum):
     """Enum for LLM models available for TEE execution"""
-    
+
     # 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 = "openai/gpt-4.1-2025-04-14"
     GPT_4O = "openai/gpt-4o"
     O4_MINI = "openai/o4-mini"
-    
+
     # Anthropic models via TEE
     CLAUDE_3_7_SONNET = "anthropic/claude-3.7-sonnet"
     CLAUDE_3_5_HAIKU = "anthropic/claude-3.5-haiku"
     CLAUDE_4_0_SONNET = "anthropic/claude-4.0-sonnet"
-    
+
     # Google models via TEE
     GEMINI_2_5_FLASH = "google/gemini-2.5-flash"
     GEMINI_2_5_PRO = "google/gemini-2.5-pro"
     GEMINI_2_0_FLASH = "google/gemini-2.0-flash"
     GEMINI_2_5_FLASH_LITE = "google/gemini-2.5-flash-lite"
-    
+
     # xAI Grok models via TEE
     GROK_3_MINI_BETA = "x-ai/grok-3-mini-beta"
     GROK_3_BETA = "x-ai/grok-3-beta"

opengradient/x402_auth.py

@@ -0,0 +1,60 @@
+import httpx
+import typing
+import logging
+
+from x402.clients.base import x402Client
+from x402.types import x402PaymentRequiredResponse, PaymentRequirements
+
+
+class X402Auth(httpx.Auth):
+    """Auth class for handling x402 payment requirements."""
+
+    def __init__(
+        self,
+        account: typing.Any,
+        max_value: typing.Optional[int] = None,
+        payment_requirements_selector: typing.Optional[
+            typing.Callable[
+                [
+                    list[PaymentRequirements],
+                    typing.Optional[str],
+                    typing.Optional[str],
+                    typing.Optional[int],
+                ],
+                PaymentRequirements,
+            ]
+        ] = None,
+    ):
+        self.x402_client = x402Client(
+            account,
+            max_value=max_value,
+            payment_requirements_selector=payment_requirements_selector,  # type: ignore
+        )
+
+    async def async_auth_flow(
+        self, request: httpx.Request
+    ) -> typing.AsyncGenerator[httpx.Request, httpx.Response]:
+        response = yield request
+
+        if response.status_code == 402:
+            try:
+                await response.aread()
+                data = response.json()
+
+                payment_response = x402PaymentRequiredResponse(**data)
+
+                selected_requirements = self.x402_client.select_payment_requirements(
+                    payment_response.accepts
+                )
+
+                payment_header = self.x402_client.create_payment_header(
+                    selected_requirements, payment_response.x402_version
+                )
+
+                request.headers["X-Payment"] = payment_header
+                request.headers["Access-Control-Expose-Headers"] = "X-Payment-Response"
+                yield request
+
+            except Exception as e:
+                logging.error(f"X402Auth: Error handling payment: {e}")
+                return

{opengradient-0.5.7.dist-info → opengradient-0.5.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: opengradient
-Version: 0.5.7
+Version: 0.5.9
 Summary: Python SDK for OpenGradient decentralized model management & inference services
 Author-email: OpenGradient <kyle@vannalabs.ai>
 License-Expression: MIT
@@ -23,8 +23,7 @@ Requires-Dist: requests>=2.32.3
 Requires-Dist: langchain>=0.3.7
 Requires-Dist: openai>=1.58.1
 Requires-Dist: pydantic>=2.9.2
-Requires-Dist: og-test-x402==0.0.1
-Requires-Dist: x402==0.2.1
+Requires-Dist: og-test-x402==0.0.9
 Dynamic: license-file
 
 # OpenGradient Python SDK
@@ -133,6 +132,10 @@ For comprehensive documentation, API reference, and examples, visit:
 - [OpenGradient Documentation](https://docs.opengradient.ai/)
 - [API Reference](https://docs.opengradient.ai/api_reference/python_sdk/)
 
+### Claude Code Users
+
+If you use [Claude Code](https://claude.ai/code), copy [docs/CLAUDE_SDK_USERS.md](docs/CLAUDE_SDK_USERS.md) to your project's `CLAUDE.md` to help Claude assist you with OpenGradient SDK development.
+
 ## Support
 
 - Run `opengradient --help` for CLI command reference

{opengradient-0.5.7.dist-info → opengradient-0.5.9.dist-info}/RECORD

@@ -1,11 +1,12 @@
-opengradient/__init__.py,sha256=wVg0KTFNBl7RnZF9huR5-m_q1E7tO-YyQwY7AD9JFoc,12635
+opengradient/__init__.py,sha256=1PSbDRGe4ft_0FYoPS3XpeajnRPOTkmx8aZZxcOeztQ,13455
 opengradient/account.py,sha256=5wrYpws_1lozjOFjLCTHtxgoxK-LmObDAaVy9eDcJY4,1145
-opengradient/cli.py,sha256=RksBEGVcZgUg6ng53Fgz-Ncv1erBwdADgblB2HmKkwk,29868
-opengradient/client.py,sha256=lzuC4rEwyf1EgHw49PMXjIyjyTsBIAP2XmCNp0cCNzg,63113
-opengradient/defaults.py,sha256=w8-dr5ciF2TGnqbm_ib0Yz4U0YL5ikpNqkcPVpmXzP8,673
+opengradient/cli.py,sha256=pfgyLfD1MIDifKmGLFsJqBlgvqIcnsIh3zzg7PaIeH4,33670
+opengradient/client.py,sha256=KDkFxcZ-vGyriFW-ydWTnitgV6rYfxtnNzchWca-8u8,74009
+opengradient/defaults.py,sha256=YOtFDq8HiwEkgMXlV4Zf3YgkopfKUkkx0CpgNuY_Mxk,796
 opengradient/exceptions.py,sha256=88tfegboGtlehQcwhxsl6ZzhLJWZWlkf_bkHTiCtXpo,3391
-opengradient/types.py,sha256=L60W1O3xmlYxbJ9ouN6nKNKJSdTW3yzCDsU0lozWF8U,7837
+opengradient/types.py,sha256=bADakUM6WwdMORGC5HvQvWCezNwIlVc7l0zodPapbhQ,14622
 opengradient/utils.py,sha256=ZUq4OBIml2vsC0tRqus4Zwb_e3g4woo00apByrafuVw,8058
+opengradient/x402_auth.py,sha256=Jmj-40OybugOXIt_qHzN1qy4x7U3QuM1MKNmPzoEKwc,1920
 opengradient/abi/InferencePrecompile.abi,sha256=reepTHg6Q01UrFP0Gexc-JayplsvOLPfG7jrEZ-cV28,10197
 opengradient/abi/PriceHistoryInference.abi,sha256=ZB3fZdx1kaFlp2wt1vTbTZZG1k8HPvmNtkG5Q8Bnajw,5098
 opengradient/abi/WorkflowScheduler.abi,sha256=yEGs76qO4S1z980KL5hBdfyXiJ6k-kERcB1O_o73AEU,416
@@ -16,7 +17,7 @@ opengradient/alphasense/run_model_tool.py,sha256=wlDqXVHa1xpqQy_hmht_wWegxtqdYgY
 opengradient/alphasense/types.py,sha256=uxk4JQKbaS2cM3ZiKpdHQb234OJ5ylprNR5vi01QFzA,220
 opengradient/bin/PriceHistoryInference.bin,sha256=nU2FZpGHIKBZ7NSK9Sr-p9lr-nXja_40ISPN9yckDq8,41276
 opengradient/llm/__init__.py,sha256=eYFBrOf1GZr0VGbIw-gSFr8hM3Rbw74ye8l-pnBPNuA,1104
-opengradient/llm/og_langchain.py,sha256=inJmu28aWA1RAWJx8G0h8w4FGXSKMCzJVOBGUwiKqnI,4904
+opengradient/llm/og_langchain.py,sha256=fVHEq_hJbWrLLVZXKSH5wwSG5kQEt_PGnmAOLUnEgmw,4965
 opengradient/llm/og_openai.py,sha256=26W_NDnLaICIaWbi9aou40v5ZJXLlmLdztDrdFoDGAU,3789
 opengradient/proto/__init__.py,sha256=AhaSmrqV0TXGzCKaoPV8-XUvqs2fGAJBM2aOmDpkNbE,55
 opengradient/proto/infer.proto,sha256=13eaEMcppxkBF8yChptsX9HooWFwJKze7oLZNl-LEb8,1217
@@ -27,9 +28,9 @@ opengradient/workflow_models/constants.py,sha256=viIkb_LGcfVprqQNaA80gBTj6cfYam0
 opengradient/workflow_models/types.py,sha256=Z22hF6c8Y4D2GlzVEIBODGwsqSjSrQvUcpZ7R-mIJdI,409
 opengradient/workflow_models/utils.py,sha256=ySfpuiOBqLTlfto6ZxZf2vc7K6RGIja0l4eaVm5AOzY,1503
 opengradient/workflow_models/workflow_models.py,sha256=d4C_gs39DAfy4cdY9Ee6GMXpPfzwvKFpmxzK1A7LNgU,3900
-opengradient-0.5.7.dist-info/licenses/LICENSE,sha256=xEcvQ3AxZOtDkrqkys2Mm6Y9diEnaSeQRKvxi-JGnNA,1069
-opengradient-0.5.7.dist-info/METADATA,sha256=WmJWW2IPbsKD_fuhafHf_cZeLiCz9_2T8qZUcvw0saU,4019
-opengradient-0.5.7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-opengradient-0.5.7.dist-info/entry_points.txt,sha256=yUKTaJx8RXnybkob0J62wVBiCp_1agVbgw9uzsmaeJc,54
-opengradient-0.5.7.dist-info/top_level.txt,sha256=oC1zimVLa2Yi1LQz8c7x-0IQm92milb5ax8gHBHwDqU,13
-opengradient-0.5.7.dist-info/RECORD,,
+opengradient-0.5.9.dist-info/licenses/LICENSE,sha256=xEcvQ3AxZOtDkrqkys2Mm6Y9diEnaSeQRKvxi-JGnNA,1069
+opengradient-0.5.9.dist-info/METADATA,sha256=kSTyBctZ-r4h3ilq7DRgxvxQhYO4ejUl3KbwcDX1Ygs,4215
+opengradient-0.5.9.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+opengradient-0.5.9.dist-info/entry_points.txt,sha256=yUKTaJx8RXnybkob0J62wVBiCp_1agVbgw9uzsmaeJc,54
+opengradient-0.5.9.dist-info/top_level.txt,sha256=oC1zimVLa2Yi1LQz8c7x-0IQm92milb5ax8gHBHwDqU,13
+opengradient-0.5.9.dist-info/RECORD,,

{opengradient-0.5.7.dist-info → opengradient-0.5.9.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.9.0)
+Generator: setuptools (80.10.2)
 Root-Is-Purelib: true
 Tag: py3-none-any