LLM-Bridge 2.2.1__tar.gz → 2.2.3__tar.gz

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: LLM-Bridge
-Version: 2.2.1
+Version: 2.2.3
 Summary: A library that wraps multiple LLM providers into a consistent API while using each provider's native SDK internally, supporting multimodal I/O, file processing, and stream output.
 Author-email: windsnow1025 <windsnow1025@gmail.com>
 License-Expression: MIT

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/llm_bridge/client/implementations/claude/non_stream_claude_client.py

@@ -14,13 +14,14 @@ class NonStreamClaudeClient(ClaudeClient):
 
             message = await self.client.beta.messages.create(
                 model=self.model,
-                max_tokens=self.max_tokens,
-                system=self.system,
                 messages=serialize(self.messages),
+                system=self.system,
+                max_tokens=self.max_tokens,
                 betas=self.betas,
                 tools=self.tools,
                 cache_control=self.cache_control,
                 thinking=self.thinking,
+                speed=self.speed,
                 output_config=self.output_config,
             )
 

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/llm_bridge/client/implementations/claude/stream_claude_client.py

@@ -34,15 +34,16 @@ class StreamClaudeClient(ClaudeClient):
 
             stream: AsyncStream[BetaRawMessageStreamEvent] = await self.client.beta.messages.create(
                 model=self.model,
-                max_tokens=self.max_tokens,
-                system=self.system,
                 messages=serialize(self.messages),
+                system=self.system,
+                max_tokens=self.max_tokens,
+                stream=True,
                 betas=self.betas,
                 tools=self.tools,
                 cache_control=self.cache_control,
                 thinking=self.thinking,
+                speed=self.speed,
                 output_config=self.output_config,
-                stream=True,
             )
         except Exception as e:
             raise_http_exception(e)

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/llm_bridge/client/model_client/claude_client.py

@@ -1,4 +1,5 @@
 from collections.abc import AsyncGenerator
+from typing import Literal
 
 import anthropic
 from anthropic import Omit
@@ -23,6 +24,7 @@ class ClaudeClient(ChatClient):
             tools: list[BetaToolUnionParam],
             cache_control: BetaCacheControlEphemeralParam,
             thinking: BetaThinkingConfigParam | Omit,
+            speed: Literal["standard", "fast"] | Omit,
             output_config: BetaOutputConfigParam | Omit,
     ):
         self.model = model
@@ -34,6 +36,7 @@ class ClaudeClient(ChatClient):
         self.tools = tools
         self.cache_control = cache_control
         self.thinking = thinking
+        self.speed = speed
         self.output_config = output_config
 
     async def generate_non_stream_response(self) -> ChatResponse:

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/llm_bridge/logic/chat_generate/model_client_factory/claude_client_factory.py

@@ -1,4 +1,4 @@
-from typing import Any
+from typing import Any, Literal
 
 import anthropic
 from anthropic import Omit, transform_schema
@@ -62,6 +62,11 @@ async def create_claude_client(
         thinking = BetaThinkingConfigAdaptiveParam(type="adaptive", display="summarized")
 
     betas: list[AnthropicBetaParam] | Omit = omit
+    speed: Literal["standard", "fast"] | Omit = omit
+    # Fast mode is a research preview with a separate rate limit; re-enable when it reaches GA
+    # if model in ["claude-opus-5-5"]:
+    #     betas = ["fast-mode-2026-02-01"]
+    #     speed = "fast"
 
     tools: list[BetaToolUnionParam] = []
     if web_search:
@@ -103,6 +108,7 @@ async def create_claude_client(
             tools=tools,
             cache_control=cache_control,
             thinking=thinking,
+            speed=speed,
             output_config=output_config,
         )
     else:
@@ -116,5 +122,6 @@ async def create_claude_client(
             tools=tools,
             cache_control=cache_control,
             thinking=thinking,
+            speed=speed,
             output_config=output_config,
         )

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/llm_bridge/resources/model_prices.json

@@ -122,8 +122,8 @@
   {
     "apiType": "Claude",
     "model": "claude-opus-5-5",
-    "input": 4,
-    "output": 20
+    "input": 8,
+    "output": 40
   },
   {
     "apiType": "Claude",

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "LLM-Bridge"
-version = "2.2.1"
+version = "2.2.3"
 dependencies = [
     "fastapi",
     "httpx",

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/uv.lock

@@ -794,7 +794,7 @@ wheels = [
 
 [[package]]
 name = "llm-bridge"
-version = "2.2.1"
+version = "2.2.3"
 source = { editable = "." }
 dependencies = [
     { name = "anthropic" },

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/llm_bridge/client/implementations/openai_completion/non_stream_openai_completion_client.py

@@ -14,8 +14,8 @@ class NonStreamOpenAICompletionClient(OpenAICompletionClient):
             logging.info(f"messages: {self.messages}")
 
             completion: ChatCompletion = await self.client.chat.completions.create(
-                messages=serialize(self.messages),
                 model=self.model,
+                messages=serialize(self.messages),
                 temperature=self.temperature,
                 stream=False,
                 reasoning_effort=self.reasoning_effort,

{llm_bridge-2.2.1 → llm_bridge-2.2.3}/llm_bridge/client/implementations/openai_completion/stream_openai_completion_client.py

@@ -61,8 +61,8 @@ class StreamOpenAICompletionClient(OpenAICompletionClient):
             logging.info(f"messages: {self.messages}")
 
             completion: AsyncStream[ChatCompletionChunk] = await self.client.chat.completions.create(
-                messages=serialize(self.messages),
                 model=self.model,
+                messages=serialize(self.messages),
                 temperature=self.temperature,
                 stream=True,
                 stream_options=ChatCompletionStreamOptionsParam(include_usage=True),