# dazllm 🚀
**Simple, unified interface for all major LLMs**
Stop juggling different APIs and libraries. `dazllm` gives you a clean, consistent way to chat with any LLM - from GPT-4 and Claude to local Ollama or LM Studio models.
## Features
✨ **Unified API** - Same interface for OpenAI, Anthropic, Google, and local models (Ollama, LM Studio)
🔧 **Smart Model Selection** - Choose by name, type, or let it auto-select
🔐 **Secure Configuration** - API keys stored safely in system keyring
📝 **Structured Output** - Get Pydantic models directly from LLM responses
🎨 **Image Generation** - Create images with DALL-E and more
💻 **CLI & Python API** - Use from command line or import in your code
## Quick Start
### Installation
```bash
pip install dazllm
```
### Setup
Configure your API keys using keyring:
```bash
keyring set dazllm openai_api_key YOUR_OPENAI_KEY
keyring set dazllm anthropic_api_key YOUR_ANTHROPIC_KEY
keyring set dazllm google_api_key YOUR_GOOGLE_KEY
keyring set dazllm ollama_url http://localhost:11434
keyring set dazllm lmstudio_url http://localhost:1234
```
Check everything is working:
```bash
dazllm --check
```
### Usage
#### Command Line
```bash
# Simple chat
dazllm chat "What's the capital of France?"
# Use specific model
dazllm chat --model openai:gpt-4 "Explain quantum computing"
# Use model type (auto-selects best available)
dazllm chat --model-type paid_best "Write a poem"
# Use provider default
dazllm chat --model openai "Tell me about AI"
# Structured output
dazllm structured "List 3 colors" --schema '{"type":"array","items":{"type":"string"}}'
# Generate images
dazllm image "a red cat wearing a hat" cat.png
# From file
dazllm chat --file prompt.txt --output response.txt
```
#### Python API
```python
from dazllm import Llm, ModelType
from pydantic import BaseModel
# Instance-based usage
llm = Llm("openai:gpt-4")
response = llm.chat("Hello!")
# Static/module-level usage
response = Llm.chat("Hello!", model="anthropic:claude-3-5-sonnet-20241022")
response = Llm.chat("Hello!", model_type=ModelType.PAID_BEST)
# Structured output with Pydantic
class ColorList(BaseModel):
colors: list[str]
result = Llm.chat_structured("List 3 colors", ColorList)
print(result.colors) # ['red', 'green', 'blue']
# Image generation
Llm.image("a sunset over mountains", "sunset.png")
# Conversation history
conversation = [
{"role": "user", "content": "Hello"},
{"role": "assistant", "content": "Hi there!"},
{"role": "user", "content": "What's your name?"}
]
response = Llm.chat(conversation, model="ollama:mistral-small")
```
## Model Types
Instead of remembering model names, use semantic types:
- `local_small` - ~1B parameter models (fast, basic)
- `local_medium` - ~7B parameter models (good balance)
- `local_large` - ~14B parameter models (best local quality)
- `paid_cheap` - Cost-effective cloud models
- `paid_best` - Highest quality cloud models
## Model Format
All models use the format `provider:model`:
- **OpenAI**: `openai:gpt-4o`, `openai:gpt-4o-mini`, `openai:dall-e-3`
- **Anthropic**: `anthropic:claude-3-5-sonnet-20241022`, `anthropic:claude-3-haiku-20240307`
- **Google**: `google:gemini-pro`, `google:gemini-flash`
- **Ollama**: `ollama:mistral-small`, `ollama:llama3:8b`, `ollama:codellama:7b`
- **LM Studio**: `lm-studio:mistral`, `lm-studio:llama3`
You can also use just the provider name (e.g., `openai`) to use that provider's default model.
## Configuration
API keys are stored securely in your system keyring:
```bash
# Set API keys
keyring set dazllm openai_api_key YOUR_OPENAI_KEY
keyring set dazllm anthropic_api_key YOUR_ANTHROPIC_KEY
keyring set dazllm google_api_key YOUR_GOOGLE_KEY
keyring set dazllm ollama_url http://localhost:11434
keyring set dazllm lmstudio_url http://localhost:1234
# Set default model (optional)
keyring set dazllm default_model openai:gpt-4o
# Check what's configured
dazllm --check
```
## Examples
### Building a Chatbot
```python
from dazllm import Llm
def chatbot():
llm = Llm.model_named("openai:gpt-4o")
conversation = []
while True:
user_input = input("You: ")
if user_input.lower() == 'quit':
break
conversation.append({"role": "user", "content": user_input})
response = llm.chat(conversation)
conversation.append({"role": "assistant", "content": response})
print(f"AI: {response}")
chatbot()
```
### Data Extraction
```python
from dazllm import Llm
from pydantic import BaseModel
class Person(BaseModel):
name: str
age: int
city: str
class People(BaseModel):
people: list[Person]
text = "John Doe, age 30, lives in New York. Jane Smith, age 25, lives in LA."
result = Llm.chat_structured(
f"Extract people info from: {text}",
People,
model="openai:gpt-4o-mini"
)
for person in result.people:
print(f"{person.name} is {person.age} years old and lives in {person.city}")
```
### Image Generation Pipeline
```python
from dazllm import Llm
# Generate image description
description = Llm.chat(
"Describe a serene mountain landscape in detail",
model_type="paid_cheap"
)
# Generate the image
image_path = Llm.image(description, "mountain.png", width=1024, height=768)
print(f"Image saved to {image_path}")
```
## Requirements
- Python 3.8+
- API keys for desired providers (OpenAI, Anthropic, Google)
- Ollama or LM Studio installed for local models
## License
MIT License
## Contributing
Contributions welcome! Please see the GitHub repository for guidelines.
---
**dazllm** - Making LLMs accessible to everyone! 🚀
Raw data
{
"_id": null,
"home_page": "https://github.com/darrenoakey/dazllm",
"name": "dazllm",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "llm, ai, openai, anthropic, claude, gemini, ollama, chatgpt, gpt-4",
"author": "Darren Oakey",
"author_email": "Darren Oakey <darren.oakey@insidemind.com.au>",
"download_url": "https://files.pythonhosted.org/packages/cb/4c/d508c51335afa940ae7a875fc32cbaf805070bbe89a9f40676af35480316/dazllm-0.4.0.tar.gz",
"platform": null,
"description": "\n\n# dazllm \ud83d\ude80\n\n**Simple, unified interface for all major LLMs**\n\nStop juggling different APIs and libraries. `dazllm` gives you a clean, consistent way to chat with any LLM - from GPT-4 and Claude to local Ollama or LM Studio models.\n\n## Features\n\n\u2728 **Unified API** - Same interface for OpenAI, Anthropic, Google, and local models (Ollama, LM Studio)\n\ud83d\udd27 **Smart Model Selection** - Choose by name, type, or let it auto-select \n\ud83d\udd10 **Secure Configuration** - API keys stored safely in system keyring \n\ud83d\udcdd **Structured Output** - Get Pydantic models directly from LLM responses \n\ud83c\udfa8 **Image Generation** - Create images with DALL-E and more \n\ud83d\udcbb **CLI & Python API** - Use from command line or import in your code \n\n## Quick Start\n\n### Installation\n\n```bash\npip install dazllm\n```\n\n### Setup\n\nConfigure your API keys using keyring:\n\n```bash\nkeyring set dazllm openai_api_key YOUR_OPENAI_KEY\nkeyring set dazllm anthropic_api_key YOUR_ANTHROPIC_KEY\nkeyring set dazllm google_api_key YOUR_GOOGLE_KEY\nkeyring set dazllm ollama_url http://localhost:11434\nkeyring set dazllm lmstudio_url http://localhost:1234\n```\n\nCheck everything is working:\n\n```bash\ndazllm --check\n```\n\n### Usage\n\n#### Command Line\n\n```bash\n# Simple chat\ndazllm chat \"What's the capital of France?\"\n\n# Use specific model \ndazllm chat --model openai:gpt-4 \"Explain quantum computing\"\n\n# Use model type (auto-selects best available)\ndazllm chat --model-type paid_best \"Write a poem\"\n\n# Use provider default\ndazllm chat --model openai \"Tell me about AI\"\n\n# Structured output\ndazllm structured \"List 3 colors\" --schema '{\"type\":\"array\",\"items\":{\"type\":\"string\"}}'\n\n# Generate images\ndazllm image \"a red cat wearing a hat\" cat.png\n\n# From file\ndazllm chat --file prompt.txt --output response.txt\n```\n\n#### Python API\n\n```python\nfrom dazllm import Llm, ModelType\nfrom pydantic import BaseModel\n\n# Instance-based usage\nllm = Llm(\"openai:gpt-4\")\nresponse = llm.chat(\"Hello!\")\n\n# Static/module-level usage\nresponse = Llm.chat(\"Hello!\", model=\"anthropic:claude-3-5-sonnet-20241022\")\nresponse = Llm.chat(\"Hello!\", model_type=ModelType.PAID_BEST)\n\n# Structured output with Pydantic\nclass ColorList(BaseModel):\n colors: list[str]\n\nresult = Llm.chat_structured(\"List 3 colors\", ColorList)\nprint(result.colors) # ['red', 'green', 'blue']\n\n# Image generation\nLlm.image(\"a sunset over mountains\", \"sunset.png\")\n\n# Conversation history\nconversation = [\n {\"role\": \"user\", \"content\": \"Hello\"},\n {\"role\": \"assistant\", \"content\": \"Hi there!\"},\n {\"role\": \"user\", \"content\": \"What's your name?\"}\n]\nresponse = Llm.chat(conversation, model=\"ollama:mistral-small\")\n```\n\n## Model Types\n\nInstead of remembering model names, use semantic types:\n\n- `local_small` - ~1B parameter models (fast, basic)\n- `local_medium` - ~7B parameter models (good balance) \n- `local_large` - ~14B parameter models (best local quality)\n- `paid_cheap` - Cost-effective cloud models\n- `paid_best` - Highest quality cloud models\n\n## Model Format\n\nAll models use the format `provider:model`:\n\n- **OpenAI**: `openai:gpt-4o`, `openai:gpt-4o-mini`, `openai:dall-e-3`\n- **Anthropic**: `anthropic:claude-3-5-sonnet-20241022`, `anthropic:claude-3-haiku-20240307`\n- **Google**: `google:gemini-pro`, `google:gemini-flash`\n- **Ollama**: `ollama:mistral-small`, `ollama:llama3:8b`, `ollama:codellama:7b`\n- **LM Studio**: `lm-studio:mistral`, `lm-studio:llama3`\n\nYou can also use just the provider name (e.g., `openai`) to use that provider's default model.\n\n## Configuration\n\nAPI keys are stored securely in your system keyring:\n\n```bash\n# Set API keys\nkeyring set dazllm openai_api_key YOUR_OPENAI_KEY\nkeyring set dazllm anthropic_api_key YOUR_ANTHROPIC_KEY\nkeyring set dazllm google_api_key YOUR_GOOGLE_KEY\nkeyring set dazllm ollama_url http://localhost:11434\nkeyring set dazllm lmstudio_url http://localhost:1234\n\n# Set default model (optional)\nkeyring set dazllm default_model openai:gpt-4o\n\n# Check what's configured\ndazllm --check\n```\n\n## Examples\n\n### Building a Chatbot\n\n```python\nfrom dazllm import Llm\n\ndef chatbot():\n llm = Llm.model_named(\"openai:gpt-4o\")\n conversation = []\n \n while True:\n user_input = input(\"You: \")\n if user_input.lower() == 'quit':\n break\n \n conversation.append({\"role\": \"user\", \"content\": user_input})\n response = llm.chat(conversation)\n conversation.append({\"role\": \"assistant\", \"content\": response})\n \n print(f\"AI: {response}\")\n\nchatbot()\n```\n\n### Data Extraction\n\n```python\nfrom dazllm import Llm\nfrom pydantic import BaseModel\n\nclass Person(BaseModel):\n name: str\n age: int\n city: str\n\nclass People(BaseModel):\n people: list[Person]\n\ntext = \"John Doe, age 30, lives in New York. Jane Smith, age 25, lives in LA.\"\n\nresult = Llm.chat_structured(\n f\"Extract people info from: {text}\",\n People,\n model=\"openai:gpt-4o-mini\"\n)\n\nfor person in result.people:\n print(f\"{person.name} is {person.age} years old and lives in {person.city}\")\n```\n\n### Image Generation Pipeline\n\n```python\nfrom dazllm import Llm\n\n# Generate image description\ndescription = Llm.chat(\n \"Describe a serene mountain landscape in detail\",\n model_type=\"paid_cheap\"\n)\n\n# Generate the image\nimage_path = Llm.image(description, \"mountain.png\", width=1024, height=768)\nprint(f\"Image saved to {image_path}\")\n```\n\n## Requirements\n\n- Python 3.8+\n- API keys for desired providers (OpenAI, Anthropic, Google)\n- Ollama or LM Studio installed for local models\n\n## License\n\nMIT License\n\n## Contributing\n\nContributions welcome! Please see the GitHub repository for guidelines.\n\n---\n\n**dazllm** - Making LLMs accessible to everyone! \ud83d\ude80\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "Simple, unified interface for all major LLMs",
"version": "0.4.0",
"project_urls": {
"Bug Reports": "https://github.com/darrenoakey/dazllm/issues",
"Documentation": "https://github.com/darrenoakey/dazllm#readme",
"Homepage": "https://github.com/darrenoakey/dazllm",
"Repository": "https://github.com/darrenoakey/dazllm.git"
},
"split_keywords": [
"llm",
" ai",
" openai",
" anthropic",
" claude",
" gemini",
" ollama",
" chatgpt",
" gpt-4"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "8f20f754282fbc88f48fdba56e86b5526fabf1f1e419d5b1364346e7ea67e6b5",
"md5": "f74b81bcd4d25caa4cf316e825fdea97",
"sha256": "0a33497f0178488a9bf952c73df679426eaf79fa3a98efdb105f690168074f2b"
},
"downloads": -1,
"filename": "dazllm-0.4.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "f74b81bcd4d25caa4cf316e825fdea97",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 30349,
"upload_time": "2025-08-03T11:07:02",
"upload_time_iso_8601": "2025-08-03T11:07:02.280018Z",
"url": "https://files.pythonhosted.org/packages/8f/20/f754282fbc88f48fdba56e86b5526fabf1f1e419d5b1364346e7ea67e6b5/dazllm-0.4.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "cb4cd508c51335afa940ae7a875fc32cbaf805070bbe89a9f40676af35480316",
"md5": "5d004aefa6cd36f5c28ac8430c3bb345",
"sha256": "95b4246cedbd3b54ce257d5bfefc35bffeb94eae157fedea53da677fc9428dac"
},
"downloads": -1,
"filename": "dazllm-0.4.0.tar.gz",
"has_sig": false,
"md5_digest": "5d004aefa6cd36f5c28ac8430c3bb345",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 523034,
"upload_time": "2025-08-03T11:07:04",
"upload_time_iso_8601": "2025-08-03T11:07:04.122533Z",
"url": "https://files.pythonhosted.org/packages/cb/4c/d508c51335afa940ae7a875fc32cbaf805070bbe89a9f40676af35480316/dazllm-0.4.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-08-03 11:07:04",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "darrenoakey",
"github_project": "dazllm",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"requirements": [
{
"name": "keyring",
"specs": [
[
">=",
"24.0.0"
]
]
},
{
"name": "requests",
"specs": [
[
">=",
"2.25.0"
]
]
},
{
"name": "colorama",
"specs": [
[
">=",
"0.4.0"
]
]
},
{
"name": "pydantic",
"specs": [
[
">=",
"2.0.0"
]
]
},
{
"name": "jsonschema",
"specs": [
[
">=",
"4.0.0"
]
]
},
{
"name": "Pillow",
"specs": [
[
">=",
"9.0.0"
]
]
},
{
"name": "openai",
"specs": [
[
">=",
"1.0.0"
]
]
},
{
"name": "anthropic",
"specs": [
[
">=",
"0.3.0"
]
]
},
{
"name": "google-generativeai",
"specs": [
[
">=",
"0.3.0"
]
]
},
{
"name": "google-ai-generativelanguage",
"specs": []
}
],
"lcname": "dazllm"
}
JSON
