| Name | LLM-Bridge JSON |
| Version |
1.7.9
JSON |
| download |
| home_page | None |
| Summary | A Bridge for LLMs |
| upload_time | 2025-07-10 13:18:12 |
| maintainer | None |
| docs_url | None |
| author | None |
| requires_python | >=3.12 |
| license | None |
| keywords |
llm
ai
|
| VCS |
|
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# LLM Bridge
LLM Bridge is a unified Python interface for interacting with LLMs, including OpenAI, OpenAI-Azure, OpenAI-GitHub, Gemini, Claude, and Grok.
GitHub: [https://github.com/windsnow1025/LLM-Bridge](https://github.com/windsnow1025/LLM-Bridge)
PyPI: [https://pypi.org/project/LLM-Bridge/](https://pypi.org/project/LLM-Bridge/)
## Workflow and Features
1. **Message Preprocessor**: extracts text content from documents (Word, Excel, PPT, Code files, PDFs) which are not natively supported by the target model.
2. **Chat Client Factory**: create a client for the specific LLM API with model parameters
1. **Model Message Converter**: convert general messages to model messages
1. **Media Processor**: converts media (Image, Audio, Video, PDF) which are natively supported by the target model into compatible formats.
3. **Chat Client**: generate stream or non-stream responses
1. **Model Thoughts**: captures and formats the model's thinking process
2. **Search Citations**: extracts and formats citations from search results
3. **Token Counter**: tracks and reports input and output token usage
### Model Features
The features listed represent the maximum capabilities of each API type, not necessarily those of every individual model.
| Model Type | Input Format | Capabilities | Output Format |
|------------|--------------------------------|----------------------|---------------|
| OpenAI | Text, Image | Thinking, Web Search | Text |
| Gemini | Text, Image, Video, Audio, PDF | Thinking, Web Search | Text, Image |
| Claude | Text, Image, PDF | Thinking, Web Search | Text |
| Grok | Text, Image | | Text |
## Installation
```bash
pip install --upgrade llm_bridge
```
## Test
```bash
pytest
```
## Quick Start
See `./usage/`
### Workflow
```python
from typing import AsyncGenerator
from llm_bridge import *
async def workflow(
api_keys: dict[str, str],
messages: list[Message],
model: str,
api_type: str,
temperature: float,
stream: bool
) -> ChatResponse | AsyncGenerator[ChatResponse, None]:
await preprocess_messages(messages, api_type)
chat_client = await create_chat_client(
messages=messages,
model=model,
api_type=api_type,
temperature=temperature,
stream=stream,
api_keys=api_keys,
)
if stream:
return chat_client.generate_stream_response()
else:
return await chat_client.generate_non_stream_response()
```
### Main
```python
import asyncio
import logging
import os
from pprint import pprint
from dotenv import load_dotenv
from llm_bridge import *
from usage.workflow import workflow
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
load_dotenv(".env")
api_keys = {
"OPENAI_API_KEY": os.environ.get("OPENAI_API_KEY"),
"AZURE_API_KEY": os.environ.get("AZURE_API_KEY"),
"AZURE_API_BASE": os.environ.get("AZURE_API_BASE"),
"GITHUB_API_KEY": os.environ.get("GITHUB_API_KEY"),
"GEMINI_FREE_API_KEY": os.environ.get("GEMINI_FREE_API_KEY"),
"GEMINI_PAID_API_KEY": os.environ.get("GEMINI_PAID_API_KEY"),
"ANTHROPIC_API_KEY": os.environ.get("ANTHROPIC_API_KEY"),
"XAI_API_KEY": os.environ.get("XAI_API_KEY"),
}
messages = [
Message(
role=Role.System,
contents=[
Content(type=ContentType.Text, data="You are a helpful assistant.")
]
),
Message(
role=Role.User,
contents=[
Content(type=ContentType.Text, data="Hello")
]
),
Message(
role=Role.Assistant,
contents=[
Content(type=ContentType.Text, data="Hello! How can I assist you today?")
]
),
Message(
role=Role.User,
contents=[
Content(type=ContentType.Text, data="Explain the concept of Occam's Razor and provide a simple, everyday example."),
# Content(type=ContentType.Text, data="What's the weather in NYC today?"),
# Content(type=ContentType.Text, data="Please generate an image of a cat."),
]
),
# Message(
# role=Role.User,
# contents=[
# # Content(type=ContentType.File, data="https://www.windsnow1025.com/minio/windsnow/uploads/1/1746208707489-image.png"),
# # Content(type=ContentType.File, data="https://www.windsnow1025.com/minio/windsnow/uploads/1/1746209841847-A%20Tutorial%20on%20Spectral%20Clustering.pdf"),
# # Content(type=ContentType.File, data="https://www.windsnow1025.com/minio/windsnow/uploads/1/1746212253473-file_example_MP3_700KB.mp3"),
# # Content(type=ContentType.File, data="https://www.windsnow1025.com/minio/windsnow/uploads/1/1746212980820-file_example_MP4_480_1_5MG.mp4"),
# Content(type=ContentType.Text, data="What's this?"),
# ]
# ),
]
# See /llm_bridge/resources/model_prices.json for available models
# model = "gpt-4.1"
# model = "gemini-2.5-flash-preview-native-audio-dialog"
# model = "gemini-2.5-pro-exp-03-25"
model = "gemini-2.5-pro-preview-05-06"
# model = "claude-sonnet-4-0"
# api_type = "OpenAI"
# api_type = "Gemini-Free"
api_type = "Gemini-Paid"
# api_type = "Claude"
temperature = 0
stream = True
async def main():
model_prices = get_model_prices()
pprint(model_prices)
input_tokens = 0
output_tokens = 0
response = await workflow(api_keys, messages, model, api_type, temperature, stream)
text = ""
if stream:
async for chunk in response:
pprint(chunk)
if chunk.text:
text += chunk.text
if chunk.input_tokens:
input_tokens = chunk.input_tokens
if chunk.output_tokens:
output_tokens += chunk.output_tokens
else:
pprint(response)
text = response.text
input_tokens = response.input_tokens
output_tokens = response.output_tokens
total_cost = calculate_chat_cost(api_type, model, input_tokens, output_tokens)
print(text)
print(f'Input tokens: {input_tokens}, Output tokens: {output_tokens}, Total cost: ${total_cost}')
if __name__ == "__main__":
asyncio.run(main())
```
Raw data
{
"_id": null,
"home_page": null,
"name": "LLM-Bridge",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.12",
"maintainer_email": null,
"keywords": "llm, ai",
"author": null,
"author_email": "windsnow1025 <windsnow1025@gmail.com>",
"download_url": "https://files.pythonhosted.org/packages/74/86/cc635175a45e8c92673807255b79394c3649c7fdaf8b3f231c3b64135c70/llm_bridge-1.7.9.tar.gz",
"platform": null,
"description": "# LLM Bridge\n\nLLM Bridge is a unified Python interface for interacting with LLMs, including OpenAI, OpenAI-Azure, OpenAI-GitHub, Gemini, Claude, and Grok.\n\nGitHub: [https://github.com/windsnow1025/LLM-Bridge](https://github.com/windsnow1025/LLM-Bridge)\n\nPyPI: [https://pypi.org/project/LLM-Bridge/](https://pypi.org/project/LLM-Bridge/)\n\n## Workflow and Features\n\n1. **Message Preprocessor**: extracts text content from documents (Word, Excel, PPT, Code files, PDFs) which are not natively supported by the target model.\n2. **Chat Client Factory**: create a client for the specific LLM API with model parameters\n 1. **Model Message Converter**: convert general messages to model messages\n 1. **Media Processor**: converts media (Image, Audio, Video, PDF) which are natively supported by the target model into compatible formats.\n3. **Chat Client**: generate stream or non-stream responses\n 1. **Model Thoughts**: captures and formats the model's thinking process\n 2. **Search Citations**: extracts and formats citations from search results\n 3. **Token Counter**: tracks and reports input and output token usage\n\n### Model Features\n\nThe features listed represent the maximum capabilities of each API type, not necessarily those of every individual model.\n\n| Model Type | Input Format | Capabilities | Output Format |\n|------------|--------------------------------|----------------------|---------------|\n| OpenAI | Text, Image | Thinking, Web Search | Text |\n| Gemini | Text, Image, Video, Audio, PDF | Thinking, Web Search | Text, Image |\n| Claude | Text, Image, PDF | Thinking, Web Search | Text |\n| Grok | Text, Image | | Text |\n\n## Installation\n\n```bash\npip install --upgrade llm_bridge\n```\n\n## Test\n\n```bash\npytest\n```\n\n## Quick Start\n\nSee `./usage/`\n\n### Workflow\n\n```python\nfrom typing import AsyncGenerator\n\nfrom llm_bridge import *\n\n\nasync def workflow(\n api_keys: dict[str, str],\n messages: list[Message],\n model: str,\n api_type: str,\n temperature: float,\n stream: bool\n) -> ChatResponse | AsyncGenerator[ChatResponse, None]:\n await preprocess_messages(messages, api_type)\n\n chat_client = await create_chat_client(\n messages=messages,\n model=model,\n api_type=api_type,\n temperature=temperature,\n stream=stream,\n api_keys=api_keys,\n )\n\n if stream:\n return chat_client.generate_stream_response()\n else:\n return await chat_client.generate_non_stream_response()\n```\n\n### Main\n\n```python\nimport asyncio\nimport logging\nimport os\nfrom pprint import pprint\n\nfrom dotenv import load_dotenv\n\nfrom llm_bridge import *\nfrom usage.workflow import workflow\n\nlogging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')\n\nload_dotenv(\".env\")\n\napi_keys = {\n \"OPENAI_API_KEY\": os.environ.get(\"OPENAI_API_KEY\"),\n \"AZURE_API_KEY\": os.environ.get(\"AZURE_API_KEY\"),\n \"AZURE_API_BASE\": os.environ.get(\"AZURE_API_BASE\"),\n \"GITHUB_API_KEY\": os.environ.get(\"GITHUB_API_KEY\"),\n \"GEMINI_FREE_API_KEY\": os.environ.get(\"GEMINI_FREE_API_KEY\"),\n \"GEMINI_PAID_API_KEY\": os.environ.get(\"GEMINI_PAID_API_KEY\"),\n \"ANTHROPIC_API_KEY\": os.environ.get(\"ANTHROPIC_API_KEY\"),\n \"XAI_API_KEY\": os.environ.get(\"XAI_API_KEY\"),\n}\n\nmessages = [\n Message(\n role=Role.System,\n contents=[\n Content(type=ContentType.Text, data=\"You are a helpful assistant.\")\n ]\n ),\n Message(\n role=Role.User,\n contents=[\n Content(type=ContentType.Text, data=\"Hello\")\n ]\n ),\n Message(\n role=Role.Assistant,\n contents=[\n Content(type=ContentType.Text, data=\"Hello! How can I assist you today?\")\n ]\n ),\n Message(\n role=Role.User,\n contents=[\n Content(type=ContentType.Text, data=\"Explain the concept of Occam's Razor and provide a simple, everyday example.\"),\n # Content(type=ContentType.Text, data=\"What's the weather in NYC today?\"),\n # Content(type=ContentType.Text, data=\"Please generate an image of a cat.\"),\n ]\n ),\n # Message(\n # role=Role.User,\n # contents=[\n # # Content(type=ContentType.File, data=\"https://www.windsnow1025.com/minio/windsnow/uploads/1/1746208707489-image.png\"),\n # # Content(type=ContentType.File, data=\"https://www.windsnow1025.com/minio/windsnow/uploads/1/1746209841847-A%20Tutorial%20on%20Spectral%20Clustering.pdf\"),\n # # Content(type=ContentType.File, data=\"https://www.windsnow1025.com/minio/windsnow/uploads/1/1746212253473-file_example_MP3_700KB.mp3\"),\n # # Content(type=ContentType.File, data=\"https://www.windsnow1025.com/minio/windsnow/uploads/1/1746212980820-file_example_MP4_480_1_5MG.mp4\"),\n # Content(type=ContentType.Text, data=\"What's this?\"),\n # ]\n # ),\n]\n# See /llm_bridge/resources/model_prices.json for available models\n# model = \"gpt-4.1\"\n# model = \"gemini-2.5-flash-preview-native-audio-dialog\"\n# model = \"gemini-2.5-pro-exp-03-25\"\nmodel = \"gemini-2.5-pro-preview-05-06\"\n# model = \"claude-sonnet-4-0\"\n# api_type = \"OpenAI\"\n# api_type = \"Gemini-Free\"\napi_type = \"Gemini-Paid\"\n# api_type = \"Claude\"\ntemperature = 0\nstream = True\n\n\nasync def main():\n model_prices = get_model_prices()\n pprint(model_prices)\n\n input_tokens = 0\n output_tokens = 0\n response = await workflow(api_keys, messages, model, api_type, temperature, stream)\n text = \"\"\n if stream:\n async for chunk in response:\n pprint(chunk)\n if chunk.text:\n text += chunk.text\n if chunk.input_tokens:\n input_tokens = chunk.input_tokens\n if chunk.output_tokens:\n output_tokens += chunk.output_tokens\n else:\n pprint(response)\n text = response.text\n input_tokens = response.input_tokens\n output_tokens = response.output_tokens\n total_cost = calculate_chat_cost(api_type, model, input_tokens, output_tokens)\n print(text)\n print(f'Input tokens: {input_tokens}, Output tokens: {output_tokens}, Total cost: ${total_cost}')\n\n\nif __name__ == \"__main__\":\n asyncio.run(main())\n```\n",
"bugtrack_url": null,
"license": null,
"summary": "A Bridge for LLMs",
"version": "1.7.9",
"project_urls": null,
"split_keywords": [
"llm",
" ai"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "c66e39c582937c97c51a6568ca91aa96cabd7de866a5486bb8805569de46f7bb",
"md5": "6c1f85a73ad5c9293acf6afed135edac",
"sha256": "1b102591696b5f8ef8f085684936b746697027d60d373742bc8614cb08acbe5f"
},
"downloads": -1,
"filename": "llm_bridge-1.7.9-py3-none-any.whl",
"has_sig": false,
"md5_digest": "6c1f85a73ad5c9293acf6afed135edac",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.12",
"size": 42012,
"upload_time": "2025-07-10T13:18:10",
"upload_time_iso_8601": "2025-07-10T13:18:10.967762Z",
"url": "https://files.pythonhosted.org/packages/c6/6e/39c582937c97c51a6568ca91aa96cabd7de866a5486bb8805569de46f7bb/llm_bridge-1.7.9-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "7486cc635175a45e8c92673807255b79394c3649c7fdaf8b3f231c3b64135c70",
"md5": "01f22045dc73d42d28c237f69e4faf73",
"sha256": "cb0e32e3da8f397df42abe4f57d998bcbea8c9f91c2e4ed1bb798d4c6f65ed6e"
},
"downloads": -1,
"filename": "llm_bridge-1.7.9.tar.gz",
"has_sig": false,
"md5_digest": "01f22045dc73d42d28c237f69e4faf73",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.12",
"size": 23051,
"upload_time": "2025-07-10T13:18:12",
"upload_time_iso_8601": "2025-07-10T13:18:12.553051Z",
"url": "https://files.pythonhosted.org/packages/74/86/cc635175a45e8c92673807255b79394c3649c7fdaf8b3f231c3b64135c70/llm_bridge-1.7.9.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-07-10 13:18:12",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "llm-bridge"
}
