openrouter

kagisearch · Oct 18, 2024 · e01a64c · e01a64c
1 parent 86c3df1
commit e01a64c
Show file tree

Hide file tree

Showing 5 changed files with 18 additions and 14 deletions.
diff --git a/llms/providers/bedrock_anthropic.py b/llms/providers/bedrock_anthropic.py
@@ -19,6 +19,7 @@ class BedrockAnthropicProvider(AnthropicProvider):
         "anthropic.claude-v2": {"prompt": 11.02, "completion": 32.68, "token_limit": 100_000},
         "anthropic.claude-3-haiku-20240307-v1:0": {"prompt": 0.25, "completion": 1.25, "token_limit": 200_000, "output_limit": 4_096},
         "anthropic.claude-3-sonnet-20240229-v1:0": {"prompt": 3.00, "completion": 15, "token_limit": 200_000, "output_limit": 4_096},
+        "anthropic.claude-3-5-sonnet-20240620-v1:0": {"prompt": 3.00, "completion": 15, "token_limit": 200_000, "output_limit": 4_096},
     }
 
     def __init__(

diff --git a/llms/providers/google_genai.py b/llms/providers/google_genai.py
@@ -4,7 +4,7 @@
 import os, math
 from typing import Dict
 
-import google.generativeai as palm
+import google.generativeai as genai
 
 from ..results.result import Result
 from .base_provider import BaseProvider
@@ -17,7 +17,10 @@ class GoogleGenAIProvider(BaseProvider):
         "chat-bison-genai": {"prompt": 0.5, "completion": 0.5, "token_limit": 0, "uses_characters": True},
         "text-bison-genai": {"prompt": 1.0, "completion": 1.0, "token_limit": 0, "uses_characters": True},
         "gemini-1.5-pro": {"prompt": 3.5, "completion": 10.5, "token_limit": 128000, "uses_characters": True},
-        "gemini-1.5-flash": {"prompt": 0.35, "completion": 1.05, "token_limit": 128000, "uses_characters": True},
+        "gemini-1.5-pro-latest": {"prompt": 3.5, "completion": 10.5, "token_limit": 128000, "uses_characters": True},
+        "gemini-1.5-flash": {"prompt": 0.075, "completion": 0.3, "token_limit": 128000, "uses_characters": True},
+        "gemini-1.5-flash-latest": {"prompt": 0.075, "completion": 0.3, "token_limit": 128000, "uses_characters": True},
+        "gemini-1.5-pro-exp-0801" : {"prompt": 3.5, "completion": 10.5, "token_limit": 128000, "uses_characters": True},
     }
 
     def __init__(self, api_key=None, model=None, **kwargs):
@@ -27,15 +30,14 @@ def __init__(self, api_key=None, model=None, **kwargs):
         if api_key is None:
             api_key = os.getenv("GOOGLE_API_KEY")
 
-        self.client = palm.configure(api_key=api_key)
+        self.client = genai.configure(api_key=api_key)
 
         self.model = model
         if model.startswith('text-'):
-            self.client = palm.generate_text
+            self.client = genai.generate_text
             self.mode = 'text'
         else:
-            self.client = palm.chat
-            self.async_client = palm.chat_async
+            self.client = genai.GenerativeModel(model)
             self.mode = 'chat'
 
 
@@ -52,8 +54,8 @@ def _prepare_model_inputs(
             messages=kwargs.pop("messages", [])
             messages=messages + [prompt]
             model_inputs = {
-                "messages": messages,
-                "temperature": temperature,
+                #"messages": messages,
+                #"temperature": temperature,
                 **kwargs,
             }
         else:
@@ -80,10 +82,10 @@ def complete(
             **kwargs,
         )
         with self.track_latency():
-            response = self.client(**model_inputs)
+            response = self.client.generate_content(prompt)
 
         if self.mode == 'chat':
-            completion = response.last
+            completion = response.text
         else:
             completion = response.result
 

diff --git a/llms/providers/openai.py b/llms/providers/openai.py
@@ -18,7 +18,7 @@ class OpenAIProvider(BaseProvider):
         "gpt-4-1106-preview": {"prompt": 10.0, "completion": 30.0, "token_limit": 128000, "is_chat": True, "output_limit": 4_096},
         "gpt-4-turbo-preview": {"prompt": 10.0, "completion": 30.0, "token_limit": 128000, "is_chat": True, "output_limit": 4_096},
         "gpt-4-turbo": {"prompt": 10.0, "completion": 30.0, "token_limit": 128000, "is_chat": True, "output_limit": 4_096},
-        "gpt-4o": {"prompt": 5.0, "completion": 15.0, "token_limit": 128000, "is_chat": True, "output_limit": 4_096},
+        "gpt-4o": {"prompt": 2.5, "completion": 10.0, "token_limit": 128000, "is_chat": True, "output_limit": 4_096},
         "gpt-4o-mini": {"prompt": 0.15, "completion": 0.60, "token_limit": 128000, "is_chat": True, "output_limit": 4_096},
         "gpt-4o-2024-08-06": {"prompt": 2.50, "completion": 10.0, "token_limit": 128000, "is_chat": True, "output_limit": 4_096},
         "o1-preview": {"prompt": 15.0, "completion": 60.0, "token_limit": 128000, "is_chat": True, "output_limit": 4_096, "use_max_completion_tokens": True},

diff --git a/requirements.txt b/requirements.txt
@@ -15,4 +15,5 @@ protobuf
 grpcio~=1.54.2
 google-generativeai
 ollama
-reka-api
+reka-api
+together
diff --git a/setup.py b/setup.py
@@ -9,8 +9,8 @@
 
 setup(
     name="pyllms",
-    version="0.5.2",
-    description="Minimal Python library to connect to LLMs (OpenAI, Anthropic, Google Palm2/Vertex, Mistral, Ollama, AI21, Cohere, Aleph-Alpha, HuggingfaceHub), with a built-in model performance benchmark.",
+    version="0.6.0",
+    description="Minimal Python library to connect to LLMs (OpenAI, Anthropic, Google,, Mistral, Reka, Groq, Together, Ollama, AI21, Cohere, Aleph-Alpha, HuggingfaceHub), with a built-in model performance benchmark.",
     long_description=long_description,
     long_description_content_type="text/markdown",
     author="Vladimir Prelovac",