[Feature]: DAY 1 SUPPORT - Gemini 1.5

[krrishdholakia]

The Feature

https://blog.google/technology/ai/google-gemini-next-generation-model-february-2024/#build-experiment

Motivation, pitch

.

Twitter / LinkedIn details

No response

[krrishdholakia]

Added the new renamed gemini pro models.

Looks like gemini-1.5 is in private preview (model names not released). Will update this ticket once they're out.

[krrishdholakia]

Model strings added to https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json

You should now be able to call this.

Vertex AI

via SDK

import litellm
litellm.vertex_project = "hardy-device-38811" # Your Project ID
litellm.vertex_location = "us-central1"  # proj location

response = litellm.completion(model="gemini-1.5-pro", messages=[{"role": "user", "content": "write code for saying hi from LiteLLM"}])

via Proxy Server

litellm_settings: 
  vertex_project: "hardy-device-38811" # Your Project ID
  vertex_location: "us-central1" # proj location

model_list: 
  - model_name: team1-gemini-pro
     litellm_params: 
       model: gemini-1.5-pro

Google AI Studio

via SDK

import litellm
litellm.vertex_project = "hardy-device-38811" # Your Project ID
litellm.vertex_location = "us-central1"  # proj location

response = litellm.completion(model="google/gemini-1.5-pro", messages=[{"role": "user", "content": "write code for saying hi from LiteLLM"}])

via Proxy Server

litellm_settings: 
  vertex_project: "hardy-device-38811" # Your Project ID
  vertex_location: "us-central1" # proj location

model_list: 
  - model_name: team1-gemini-pro
     litellm_params: 
       model: gemini/gemini-1.5-pro

[haseeb-heaven]

Hi i tried to run Gemini 1.5 pro but i am getting this error

Here is the code

import os
import litellm # Main libray for LLM's
from dotenv import load_dotenv

def init_api_keys():
    load_dotenv()
    gemini_api_key = os.getenv("GEMINI_API_KEY")
    if not gemini_api_key:
        print("GEMINI_API_KEY not found in.env file")
        exit()

def get_model_message(prompt: str):
    system_message = "You are an intelligent coding assistant. You can generate code effeciently. \n"
    messages = [
        {"role": "system", "content":system_message},
        {"role": "assistant", "content": "Please generate code wrapped inside triple backticks known as codeblock."},
        {"role": "user", "content": prompt}
    ]
    return messages

def extract_content(output:str):
    try:
        return output['choices'][0]['message']['content']
    except (KeyError, TypeError) as exception:
        print(f"Error extracting content: {str(exception)}")
        raise

def main():
    try:
        #litellm.set_verbose=True
        # call the init_api_keys() function to load the API key
        init_api_keys()
        
        # set the model
        model_name = "gemini/gemini-1.5-pro"
        # Initialize the interpreter
        prompt = input("Enter your query: ")
        
        messages:str = get_model_message(prompt)
        temperature:float = 0.1
        
        response = litellm.completion(model_name, messages=messages,temperature=temperature)
        if not response:
            print("Error in generating response. Please try again.")
            return
        
        content = extract_content(response)
        
        if not content:
            print("Error in extracting content. Please try again.")
            return
        
        print(content)
    except Exception as exception:
        print(f"Error in main: {str(exception)}")
        
if __name__ == "__main__":
    main()

Error generating response: 404 models/gemini-1.5-pro-latest is not found for API version v1beta, or is not supported for GenerateContent. Call ListModels to see the list of available models and their supported methods.

[haseeb-heaven]

I have added new Feature Request as well for this (google-gemini/generative-ai-python#227) and wait it for to approve to get access via API.

[haseeb-heaven]

Hi @krrishdholakia The Gemini Pro 1.5 via API is still not working.

Request to litellm:
litellm.completion('gemini/gemini-1.5-pro', messages=[{'role': 'system', 'content': 'You are an intelligent coding assistant. You can generate code effeciently. \n'}, {'role': 'assistant', 'content': 'Please generate code wrapped inside triple backticks known as codeblock.'}, {'role': 'user', 'content': ' Write factorial of number in C++ 20'}], temperature=0.1)


self.optional_params: {}
kwargs[caching]: False; litellm.cache: None
Final returned optional params: {'temperature': 0.1}
self.optional_params: {'temperature': 0.1}
{'model': 'gemini-1.5-pro', 'messages': [{'role': 'system', 'content': 'You are an intelligent coding assistant. You can generate code effeciently. \n'}, {'role': 'assistant', 'content': 'Please generate code wrapped inside triple backticks known as codeblock.'}, {'role': 'user', 'content': ' Write factorial of number in C++ 20'}], 'optional_params': {'temperature': 0.1}, 'litellm_params': {'acompletion': False, 'api_key': None, 'force_timeout': 600, 'logger_fn': None, 'verbose': False, 'custom_llm_provider': 'gemini', 'api_base': '', 'litellm_call_id': '5edf1598-7f74-45f5-ab8a-a9c7ba572be3', 'model_alias_map': {}, 'completion_call_id': None, 'metadata': None, 'model_info': None, 'proxy_server_request': None, 'preset_cache_key': None, 'no-log': False, 'stream_response': {}}, 'start_time': datetime.datetime(2024, 4, 10, 0, 57, 27, 658729), 'stream': False, 'user': None, 'call_type': 'completion', 'litellm_call_id': '5edf1598-7f74-45f5-ab8a-a9c7ba572be3', 'completion_start_time': None, 'temperature': 0.1, 'input': ['You are an intelligent coding assistant. You can generate code effeciently. \nPlease generate code wrapped inside triple backticks known as codeblock. Write factorial of number in C++ 20'], 'api_key': '', 'additional_args': {'complete_input_dict': {'inference_params': {'temperature': 0.1}}}, 'log_event_type': 'pre_api_call'}


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.

Logging Details: logger_fn - None | callable(logger_fn) - False
Logging Details LiteLLM-Failure Call
self.failure_callback: []
Error in main: 404 models/gemini-1.5-pro is not found for API version v1beta, or is not supported for GenerateContent. Call ListModels to see the list of available models and their supported methods.

LiteLLM: Current Version = 1.34.38

[haseeb-heaven]

Gemini 1.5 Pro is now available through the API

https://developers.googleblog.com/2024/04/gemini-15-pro-in-public-preview-with-new-features.html?m=1

[krrishdholakia]

Hey @haseeb-heaven just checked on google ai studio - it doesn't look like it's out yet via api. Let me know if your portal shows you any different.

Error in main: 404 models/gemini-1.5-pro is not found for API version v1beta, or is not supported for GenerateContent. Call ListModels to see the list of available models and their supported methods.

^ this error looks like it's being raised by google

[krrishdholakia]

We already have gemini 1.5 support on vertex ai -

litellm/model_prices_and_context_window.json

Line 973 in dbbf605

"gemini-1.5-pro": {

and people using it there. would recommend trying to see if you have access on there.

[haseeb-heaven]

Hey @haseeb-heaven just checked on google ai studio - it doesn't look like it's out yet via api. Let me know if your portal shows you any different.

Error in main: 404 models/gemini-1.5-pro is not found for API version v1beta, or is not supported for GenerateContent. Call ListModels to see the list of available models and their supported methods.

^ this error looks like it's being raised by google

I have tried with Google generative AI python sdk and Gemini 1.5 Pro is working fine there