llm-api-adapter 0.3.1__tar.gz → 0.3.2__tar.gz

{llm_api_adapter-0.3.1 → llm_api_adapter-0.3.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llm-api-adapter
-Version: 0.3.1
+Version: 0.3.2
 Summary: Lightweight, pluggable adapter for multiple LLM APIs (OpenAI, Anthropic, Google)
 Author: Sergey Inozemtsev
 License: MIT License
@@ -41,6 +41,7 @@ Dynamic: license-file
 # LLM API Adapter SDK for Python
 [![PyPI Downloads](https://static.pepy.tech/personalized-badge/llm-api-adapter?period=total&units=INTERNATIONAL_SYSTEM&left_color=BLACK&right_color=GREEN&left_text=downloads)](https://pepy.tech/projects/llm-api-adapter)
 [![Tests](https://github.com/Inozem/llm_api_adapter/actions/workflows/ci-main.yml/badge.svg)](https://github.com/Inozem/llm_api_adapter/actions/workflows/ci-main.yml)
+[![codecov](https://codecov.io/github/Inozem/llm_api_adapter/graph/badge.svg?token=T83ZPH1F7Z)](https://codecov.io/github/Inozem/llm_api_adapter)
 
 ## Overview
 
@@ -50,7 +51,7 @@ Currently, the project supports OpenAI, Anthropic, and Google with a consistent
 
 ### Version
 
-Current version: 0.3.1
+Current version: 0.3.2
 
 
 ## Features
@@ -183,9 +184,9 @@ The SDK provides a set of standardized errors for easier debugging and integrati
 
 The SDK allows you to easily switch between LLM providers and specify the model you want to use. Currently supported providers are OpenAI, Anthropic, and Google.
 
-- **OpenAI**: You can use models like `gpt-5.4`, `gpt-5.4-mini`, `gpt-5.4-nano`, `gpt-5.2`, `gpt-5.1`, `gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `gpt-4.1`, `gpt-4.1-mini`, `gpt-4.1-nano`, `gpt-4o`, `gpt-4o-mini`.
-- **Anthropic**: Available models include `claude-opus-4-6`, `claude-sonnet-4-6`, `claude-opus-4-5`, `claude-sonnet-4-5`, `claude-haiku-4-5`, `claude-opus-4-1`, `claude-opus-4-0`, `claude-sonnet-4-0`, `claude-3-haiku-20240307`.
-- **Google**: Models such as `gemini-3.1-pro-preview`, `gemini-3.1-flash-lite-preview`, `gemini-3-flash-preview`, `gemini-2.5-pro`, `gemini-2.5-flash`, `gemini-2.5-flash-lite` can be used.
+- **OpenAI**: You can use models like `gpt-5.5`, `gpt-5.4`, `gpt-5.4-mini`, `gpt-5.4-nano`, `gpt-5.2`, `gpt-5.1`, `gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `gpt-4.1`, `gpt-4.1-mini`, `gpt-4.1-nano`, `gpt-4o`, `gpt-4o-mini`.
+- **Anthropic**: Available models include `claude-opus-4-8`, `claude-opus-4-7`, `claude-opus-4-6`, `claude-sonnet-4-6`, `claude-opus-4-5`, `claude-sonnet-4-5`, `claude-haiku-4-5`, `claude-opus-4-1`.
+- **Google**: Models such as `gemini-3.5-flash`, `gemini-3.1-pro-preview`, `gemini-3.1-flash-lite`, `gemini-3-flash-preview`, `gemini-2.5-pro`, `gemini-2.5-flash` can be used.
 
 Example:
 

{llm_api_adapter-0.3.1 → llm_api_adapter-0.3.2}/README.md

@@ -1,6 +1,7 @@
 # LLM API Adapter SDK for Python
 [![PyPI Downloads](https://static.pepy.tech/personalized-badge/llm-api-adapter?period=total&units=INTERNATIONAL_SYSTEM&left_color=BLACK&right_color=GREEN&left_text=downloads)](https://pepy.tech/projects/llm-api-adapter)
 [![Tests](https://github.com/Inozem/llm_api_adapter/actions/workflows/ci-main.yml/badge.svg)](https://github.com/Inozem/llm_api_adapter/actions/workflows/ci-main.yml)
+[![codecov](https://codecov.io/github/Inozem/llm_api_adapter/graph/badge.svg?token=T83ZPH1F7Z)](https://codecov.io/github/Inozem/llm_api_adapter)
 
 ## Overview
 
@@ -10,7 +11,7 @@ Currently, the project supports OpenAI, Anthropic, and Google with a consistent
 
 ### Version
 
-Current version: 0.3.1
+Current version: 0.3.2
 
 
 ## Features
@@ -143,9 +144,9 @@ The SDK provides a set of standardized errors for easier debugging and integrati
 
 The SDK allows you to easily switch between LLM providers and specify the model you want to use. Currently supported providers are OpenAI, Anthropic, and Google.
 
-- **OpenAI**: You can use models like `gpt-5.4`, `gpt-5.4-mini`, `gpt-5.4-nano`, `gpt-5.2`, `gpt-5.1`, `gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `gpt-4.1`, `gpt-4.1-mini`, `gpt-4.1-nano`, `gpt-4o`, `gpt-4o-mini`.
-- **Anthropic**: Available models include `claude-opus-4-6`, `claude-sonnet-4-6`, `claude-opus-4-5`, `claude-sonnet-4-5`, `claude-haiku-4-5`, `claude-opus-4-1`, `claude-opus-4-0`, `claude-sonnet-4-0`, `claude-3-haiku-20240307`.
-- **Google**: Models such as `gemini-3.1-pro-preview`, `gemini-3.1-flash-lite-preview`, `gemini-3-flash-preview`, `gemini-2.5-pro`, `gemini-2.5-flash`, `gemini-2.5-flash-lite` can be used.
+- **OpenAI**: You can use models like `gpt-5.5`, `gpt-5.4`, `gpt-5.4-mini`, `gpt-5.4-nano`, `gpt-5.2`, `gpt-5.1`, `gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `gpt-4.1`, `gpt-4.1-mini`, `gpt-4.1-nano`, `gpt-4o`, `gpt-4o-mini`.
+- **Anthropic**: Available models include `claude-opus-4-8`, `claude-opus-4-7`, `claude-opus-4-6`, `claude-sonnet-4-6`, `claude-opus-4-5`, `claude-sonnet-4-5`, `claude-haiku-4-5`, `claude-opus-4-1`.
+- **Google**: Models such as `gemini-3.5-flash`, `gemini-3.1-pro-preview`, `gemini-3.1-flash-lite`, `gemini-3-flash-preview`, `gemini-2.5-pro`, `gemini-2.5-flash` can be used.
 
 Example:
 

{llm_api_adapter-0.3.1 → llm_api_adapter-0.3.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "llm-api-adapter"
-version = "0.3.1"
+version = "0.3.2"
 description = "Lightweight, pluggable adapter for multiple LLM APIs (OpenAI, Anthropic, Google)"
 readme = "README.md"
 requires-python = ">=3.9"

{llm_api_adapter-0.3.1 → llm_api_adapter-0.3.2}/src/llm_api_adapter/adapters/anthropic_adapter.py

@@ -34,18 +34,8 @@ class AnthropicAdapter(LLMAdapterBase):
         parallel_tool_calls: Optional[bool] = None,
         previous_response: Optional[ChatResponse] = None,
     ) -> ChatResponse:
-        temperature = self._validate_parameter(
-            name="temperature",
-            value=temperature,
-            min_value=0,
-            max_value=2,
-        )
-        top_p = self._validate_parameter(
-            name="top_p",
-            value=top_p,
-            min_value=0,
-            max_value=1,
-        )
+        temperature = self._validate_parameter("temperature", temperature, 0, 2)
+        top_p = self._validate_parameter("top_p", top_p, 0, 1)
         try:
             self._validate_tools(tools)
             validated_tools = tools
@@ -87,10 +77,11 @@ class AnthropicAdapter(LLMAdapterBase):
                         reasoning_level=reasoning_level,
                         normalized_reasoning_level=normalized_reasoning_level,
                     )
-                    params["thinking"] = {
-                        "type": "enabled",
-                        "budget_tokens": normalized_reasoning_level,
-                    }
+                    params["budget_tokens"] = normalized_reasoning_level
+                if self.is_reasoning:
+                    effort = self._reasoning_level_to_effort(reasoning_level)
+                    if effort:
+                        params["effort"] = effort
             params = {k: v for k, v in params.items() if v is not None}
             _ = previous_response
             client = ClaudeSyncClient(api_key=self.api_key)
@@ -130,42 +121,60 @@ class AnthropicAdapter(LLMAdapterBase):
             return {"type": tool_choice}
         return {"type": "tool", "name": tool_choice}
 
+    def _validate_not_bool(self, level) -> None:
+        if isinstance(level, bool):
+            raise ValueError("Invalid type for level: bool is not accepted")
+
+    def _validate_reasoning_str(self, level: str) -> None:
+        if level not in self.reasoning_levels:
+            raise ValueError(
+                f"Unknown reasoning level key: {level!r}. "
+                f"Valid keys: {list(self.reasoning_levels.keys())}"
+            )
+
+    def _resolve_reasoning_int(self, level: int) -> int:
+        if isinstance(level, int):
+            return level
+        raise ValueError(
+            "Invalid type for level: expected int or str, "
+            f"got {type(level).__name__!r}"
+        )
+
+    def _reasoning_level_to_effort(self, level: str | int) -> str | None:
+        self._validate_not_bool(level)
+        if isinstance(level, str):
+            self._validate_reasoning_str(level)
+            return None if level == "none" else level
+        numeric = self._resolve_reasoning_int(level)
+        for key, threshold in self.reasoning_levels.items():
+            if threshold > 0 and numeric <= threshold:
+                return key
+        return list(self.reasoning_levels)[-1]
+
     def _normalize_reasoning_level(self, level: str | int) -> int | None:
         minimum_level = 1024
-        normalized_level: int | None = None
-        if level is not None and not self.is_reasoning:
+        if not self.is_reasoning:
             warning_message = (
                 f"Model '{self.model}' does not support reasoning — reasoning disabled."
             )
             warnings.warn(warning_message, UserWarning)
             logger.info(warning_message)
             return None
-        if isinstance(level, bool):
-            raise ValueError("Invalid type for level: bool is not accepted")
+        self._validate_not_bool(level)
         if isinstance(level, str):
-            if level in self.reasoning_levels:
-                normalized_level = self.reasoning_levels[level]
-            else:
-                raise ValueError(
-                    f"Unknown reasoning level key: {level!r}. "
-                    f"Valid keys: {list(self.reasoning_levels.keys())}"
-                )
-        if isinstance(level, int):
-            normalized_level = level
-        if normalized_level is not None:
-            if normalized_level >= minimum_level:
-                return normalized_level
-            warning_message = (
-                f"Reasoning level '{level}' is below the minimum supported value "
-                f"{minimum_level}; using {minimum_level} instead."
-            )
-            warnings.warn(warning_message, UserWarning)
-            logger.info(warning_message)
-            return minimum_level
-        raise ValueError(
-            "Invalid type for level: expected int or str, "
-            f"got {type(level).__name__!r}"
+            self._validate_reasoning_str(level)
+            numeric = self.reasoning_levels[level]
+        else:
+            numeric = self._resolve_reasoning_int(level)
+        if numeric >= minimum_level:
+            return numeric
+        warning_message = (
+            f"Reasoning level '{level}' is below the minimum supported value "
+            f"{minimum_level}; using {minimum_level} instead."
         )
+        warnings.warn(warning_message, UserWarning)
+        logger.info(warning_message)
+        return minimum_level
 
     def validate_reasoning_and_tokens(
         self,

{llm_api_adapter-0.3.1 → llm_api_adapter-0.3.2}/src/llm_api_adapter/llm_registry/llm_registry.json

@@ -1,10 +1,14 @@
 {
   "schema_version": 6,
-  "effective_date": "2026-03-19",
+  "effective_date": "2026-06-22",
   "providers": {
     "openai": {
       "currency": "USD",
       "models": {
+        "gpt-5.5": {
+          "pricing": {"in_per_1m": 5.0, "out_per_1m": 30.0},
+          "is_reasoning": true
+        },
         "gpt-5.4": {
           "pricing": {"in_per_1m": 2.5, "out_per_1m": 15.0},
           "is_reasoning": true
@@ -47,6 +51,14 @@
     "anthropic": {
       "currency": "USD",
       "models": {
+        "claude-opus-4-8": {
+          "pricing": {"in_per_1m": 5.0, "out_per_1m": 25.0},
+          "is_reasoning": true
+        },
+        "claude-opus-4-7": {
+          "pricing": {"in_per_1m": 5.0, "out_per_1m": 25.0},
+          "is_reasoning": true
+        },
         "claude-opus-4-6": {
           "pricing": {"in_per_1m": 5.0, "out_per_1m": 25.0},
           "is_reasoning": true
@@ -70,26 +82,21 @@
         "claude-opus-4-1": {
           "pricing": {"in_per_1m": 15.0, "out_per_1m": 75.0},
           "is_reasoning": true
-        },
-        "claude-opus-4-0": {
-          "pricing": {"in_per_1m": 15.0, "out_per_1m": 75.0},
-          "is_reasoning": true
-        },
-        "claude-sonnet-4-0": {
-          "pricing": {"in_per_1m": 3.0, "out_per_1m": 15.0},
-          "is_reasoning": true
-        },
-        "claude-3-haiku-20240307": {"pricing": {"in_per_1m": 0.25, "out_per_1m": 1.25}}
+        }
       }
     },
     "google": {
       "currency": "USD",
       "models": {
+        "gemini-3.5-flash": {
+          "pricing": {"in_per_1m": 1.5, "out_per_1m": 9.0},
+          "is_reasoning": true
+        },
         "gemini-3.1-pro-preview": {
           "pricing": {"in_per_1m": 2.0, "out_per_1m": 12.0},
           "is_reasoning": true
         },
-        "gemini-3.1-flash-lite-preview": {
+        "gemini-3.1-flash-lite": {
           "pricing": {"in_per_1m": 0.25, "out_per_1m": 1.5},
           "is_reasoning": true
         },
@@ -104,10 +111,6 @@
         "gemini-2.5-flash": {
           "pricing": {"in_per_1m": 0.3, "out_per_1m": 2.5},
           "is_reasoning": true
-        },
-        "gemini-2.5-flash-lite": {
-          "pricing": {"in_per_1m": 0.1, "out_per_1m": 0.4},
-          "is_reasoning": true
         }
       }
     }

{llm_api_adapter-0.3.1 → llm_api_adapter-0.3.2}/src/llm_api_adapter/llms/anthropic/sync_client.py

@@ -14,6 +14,8 @@ from ...errors.llm_api_error import (
 
 logger = logging.getLogger(__name__)
 
+_ADAPTIVE_THINKING_MODELS = ("claude-opus-4-6", "claude-sonnet-4-6", "claude-opus-4-7", "claude-opus-4-8")
+
 @dataclass
 class ClaudeSyncClient:
     api_key: str
@@ -34,11 +36,20 @@ class ClaudeSyncClient:
         return response.json()
 
     def _prepare_chat_payload_for_model(self, model: str, kwargs: dict) -> dict:
-        if model.startswith(
-            ("claude-sonnet-4-5", "claude-opus-4-1", "claude-haiku-4-5", "claude-opus-4-5",
-             "claude-opus-4-6", "claude-sonnet-4-6")
-        ):
+        budget_tokens = kwargs.pop("budget_tokens", None)
+        effort = kwargs.pop("effort", None)
+        if model.startswith(_ADAPTIVE_THINKING_MODELS):
             kwargs.pop("top_p", None)
+            if effort:
+                kwargs["thinking"] = {"type": "adaptive"}
+                kwargs["output_config"] = {"effort": effort}
+        else:
+            if model.startswith(
+                ("claude-sonnet-4-5", "claude-opus-4-1", "claude-haiku-4-5", "claude-opus-4-5")
+            ):
+                kwargs.pop("top_p", None)
+            if budget_tokens:
+                kwargs["thinking"] = {"type": "enabled", "budget_tokens": budget_tokens}
         return {"model": model, **kwargs}
 
     def _send_request(self, url: str, payload: dict, timeout_s: float | None = None):