abstractcore 2.4.2__tar.gz → 2.4.4__tar.gz

{abstractcore-2.4.2 → abstractcore-2.4.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: abstractcore
-Version: 2.4.2
+Version: 2.4.4
 Summary: Unified interface to all LLM providers with essential infrastructure for tool calling, streaming, and model management
 Author-email: Laurent-Philippe Albou <contact@abstractcore.ai>
 Maintainer-email: Laurent-Philippe Albou <contact@abstractcore.ai>
@@ -46,6 +46,11 @@ Provides-Extra: embeddings
 Requires-Dist: sentence-transformers<4.0.0,>=2.7.0; extra == "embeddings"
 Requires-Dist: numpy<2.0.0,>=1.20.0; extra == "embeddings"
 Provides-Extra: processing
+Provides-Extra: media
+Requires-Dist: Pillow<12.0.0,>=10.0.0; extra == "media"
+Requires-Dist: pymupdf4llm<1.0.0,>=0.0.20; extra == "media"
+Requires-Dist: unstructured[office]<1.0.0,>=0.10.0; extra == "media"
+Requires-Dist: pandas<3.0.0,>=1.0.0; extra == "media"
 Provides-Extra: api-providers
 Requires-Dist: abstractcore[anthropic,openai]; extra == "api-providers"
 Provides-Extra: local-providers
@@ -55,9 +60,9 @@ Requires-Dist: abstractcore[huggingface]; extra == "heavy-providers"
 Provides-Extra: all-providers
 Requires-Dist: abstractcore[anthropic,embeddings,huggingface,lmstudio,mlx,ollama,openai]; extra == "all-providers"
 Provides-Extra: all
-Requires-Dist: abstractcore[anthropic,dev,docs,embeddings,huggingface,lmstudio,mlx,ollama,openai,processing,server,test]; extra == "all"
+Requires-Dist: abstractcore[anthropic,dev,docs,embeddings,huggingface,lmstudio,media,mlx,ollama,openai,processing,server,test]; extra == "all"
 Provides-Extra: lightweight
-Requires-Dist: abstractcore[anthropic,embeddings,lmstudio,ollama,openai,processing,server]; extra == "lightweight"
+Requires-Dist: abstractcore[anthropic,embeddings,lmstudio,media,ollama,openai,processing,server]; extra == "lightweight"
 Provides-Extra: dev
 Requires-Dist: pytest>=7.0.0; extra == "dev"
 Requires-Dist: pytest-asyncio>=0.21.0; extra == "dev"
@@ -89,6 +94,11 @@ Dynamic: license-file
 
 # AbstractCore
 
+[![PyPI version](https://img.shields.io/pypi/v/abstractcore.svg)](https://pypi.org/project/abstractcore/)
+[![Python Version](https://img.shields.io/pypi/pyversions/abstractcore)](https://pypi.org/project/abstractcore/)
+[![license](https://img.shields.io/github/license/lpalbou/abstractcore)](https://github.com/lpalbou/abstractcore/blob/main/LICENSE)
+[![GitHub stars](https://img.shields.io/github/stars/lpalbou/abstractcore?style=social)](https://github.com/lpalbou/abstractcore/stargazers)
+
 A unified Python library for interaction with multiple Large Language Model (LLM) providers.
 
 **Write once, run everywhere.**
@@ -153,15 +163,68 @@ loaded_session = BasicSession.load('conversation.json', provider=llm)
 
 [Learn more about Session](docs/session.md)
 
+### Media Handling
+
+AbstractCore provides **unified media handling** across all providers with **automatic maximum resolution optimization** for best results. Upload images, PDFs, and documents using the same simple API regardless of your provider.
+
+```python
+from abstractcore import create_llm
+
+# Vision analysis - works with any vision model
+# Images automatically processed at maximum supported resolution
+llm = create_llm("openai", model="gpt-4o")
+response = llm.generate(
+    "What's in this image?",
+    media=["photo.jpg"]  # Auto-resized to model's maximum capability
+)
+
+# Document analysis - works with any model
+llm = create_llm("anthropic", model="claude-3.5-sonnet")
+response = llm.generate(
+    "Summarize this research paper",
+    media=["research_paper.pdf"]
+)
+
+# Multiple files - mix images, PDFs, spreadsheets
+response = llm.generate(
+    "Analyze these business documents",
+    media=["report.pdf", "chart.png", "data.xlsx"]
+)
+
+# Same code works with local models
+llm = create_llm("ollama", model="qwen3-vl:8b")
+response = llm.generate(
+    "Describe this screenshot",
+    media=["screenshot.png"]  # Auto-optimized for qwen3-vl
+)
+```
+
+**Key Features:**
+- **Smart Resolution**: Automatically uses maximum resolution supported by each model
+- **Format Support**: PNG, JPEG, GIF, WEBP, BMP, TIFF images; PDF, TXT, MD, CSV, TSV, JSON documents
+- **Office Documents**: DOCX, XLSX, PPT (with `pip install abstractcore[all]`)
+- **Vision Optimization**: Model-specific image processing for best vision results
+
+**Provider compatibility:**
+- **High-resolution vision**: GPT-4o (up to 4096x4096), Claude 3.5 Sonnet (up to 1568x1568)
+- **Local models**: qwen3-vl (up to 3584x3584), gemma3:4b, llama3.2-vision
+- **All models**: Automatic text extraction for non-vision models
+
+[Learn more about Media Handling](docs/media-handling-system.md)
+
 ## Key Features
 
 - **Provider Agnostic**: Seamlessly switch between OpenAI, Anthropic, Ollama, LMStudio, MLX, HuggingFace
+- **Centralized Configuration**: Global defaults and app-specific preferences at `~/.abstractcore/config/abstractcore.json`
+- **Intelligent Media Handling**: Upload images, PDFs, and documents with automatic maximum resolution optimization
+- **Vision Model Support**: Smart image processing at each model's maximum capability for best results
+- **Document Processing**: Advanced PDF extraction (PyMuPDF4LLM), Office documents (DOCX/XLSX/PPT), CSV/TSV analysis
 - **Unified Tools**: Consistent tool calling across all providers
 - **Session Management**: Persistent conversations with metadata, analytics, and complete serialization
 - **Structured Responses**: Clean, predictable output formats with Pydantic
 - **Streaming Support**: Real-time token generation for interactive experiences
 - **Embeddings**: Built-in support for semantic search and RAG applications
-- **Universal Server**: Optional OpenAI-compatible API server
+- **Universal Server**: Optional OpenAI-compatible API server with `/v1/responses` endpoint
 
 ## Supported Providers
 
@@ -199,12 +262,15 @@ response = client.chat.completions.create(
 ```
 
 **Server Features:**
-- OpenAI-compatible REST endpoints (`/v1/chat/completions`, `/v1/embeddings`, etc.)
+- OpenAI-compatible REST endpoints (`/v1/chat/completions`, `/v1/embeddings`, `/v1/responses`)
+- **NEW in v2.5.0**: OpenAI Responses API (`/v1/responses`) with native `input_file` support
 - Multi-provider support through one HTTP API
+- Comprehensive media processing (images, PDFs, Office documents, CSV/TSV)
 - Agentic CLI integration (Codex, Crush, Gemini CLI)
-- Streaming responses
+- Streaming responses with optional opt-in
 - Tool call format conversion
-- Interactive API docs at `/docs`
+- Enhanced debug logging with `--debug` flag
+- Interactive API docs at `/docs` (Swagger UI)
 
 **When to use the server:**
 - Integrating with existing OpenAI-compatible tools
@@ -365,6 +431,72 @@ Each application has comprehensive documentation with examples and advanced usag
 - Integration with shell scripts and automation
 - Standardized text processing tasks
 
+## Configuration
+
+AbstractCore provides a **centralized configuration system** that manages default models, cache directories, and logging settings from a single location. This eliminates the need to specify `--provider` and `--model` parameters repeatedly.
+
+### Quick Setup
+
+```bash
+# Check current configuration (shows how to change each setting)
+abstractcore --status
+
+# Set defaults for all applications
+abstractcore --set-global-default ollama/llama3:8b
+
+# Or configure specific applications (examples of customization)
+abstractcore --set-app-default summarizer openai gpt-4o-mini
+abstractcore --set-app-default extractor ollama qwen3:4b-instruct
+abstractcore --set-app-default judge anthropic claude-3-5-haiku
+
+# Configure logging (common examples)
+abstractcore --set-console-log-level WARNING  # Reduce console output
+abstractcore --set-console-log-level NONE     # Disable console logging
+abstractcore --enable-file-logging            # Save logs to files
+abstractcore --enable-debug-logging           # Full debug mode
+
+# Set API keys as needed
+abstractcore --set-api-key openai sk-your-key-here
+abstractcore --set-api-key anthropic your-anthropic-key
+
+# Verify configuration (includes change commands for each setting)
+abstractcore --status
+```
+
+### Priority System
+
+AbstractCore uses a clear priority system where explicit parameters always override defaults:
+
+1. **Explicit parameters** (highest priority): `summarizer doc.txt --provider openai --model gpt-4o-mini`
+2. **App-specific config**: `abstractcore --set-app-default summarizer openai gpt-4o-mini`
+3. **Global config**: `abstractcore --set-global-default openai/gpt-4o-mini`
+4. **Built-in defaults** (lowest priority): `huggingface/unsloth/Qwen3-4B-Instruct-2507-GGUF`
+
+### Usage After Configuration
+
+Once configured, apps use your defaults automatically:
+
+```bash
+# Before configuration (requires explicit parameters)
+summarizer document.pdf --provider openai --model gpt-4o-mini
+
+# After configuration (uses configured defaults)
+summarizer document.pdf
+
+# Explicit parameters still override when needed
+summarizer document.pdf --provider anthropic --model claude-3-5-sonnet
+```
+
+### Configuration Features
+
+- **Application defaults**: Different optimal models for each app
+- **Cache directories**: Configurable cache locations for models and data
+- **Logging control**: Package-wide logging levels and debug mode
+- **API key management**: Centralized API key storage
+- **Interactive setup**: `abstractcore --configure` for guided configuration
+
+**Complete guide**: [Centralized Configuration](docs/centralized-config.md)
+
 ## Documentation
 
 **📚 Complete Documentation:** [docs/](docs/) - Full documentation index and navigation guide
@@ -376,6 +508,7 @@ Each application has comprehensive documentation with examples and advanced usag
 
 ### Core Library (Python)
 - **[Python API Reference](docs/api-reference.md)** - Complete Python API documentation
+- **[Media Handling System](docs/media-handling-system.md)** - Images, PDFs, and document processing across all providers
 - **[Session Management](docs/session.md)** - Persistent conversations, serialization, and analytics
 - **[Embeddings Guide](docs/embeddings.md)** - Semantic search, RAG, and vector embeddings
 - **[Code Examples](examples/)** - Working examples for all features
@@ -401,7 +534,44 @@ for provider in providers:
     response = llm.generate("Hello!")
 ```
 
-### 2. Local Development, Cloud Production
+### 2. Vision Analysis Across Providers
+
+```python
+# Same image analysis works with any vision model
+image_files = ["product_photo.jpg", "user_feedback.png"]
+prompt = "Analyze these product images and suggest improvements"
+
+# OpenAI GPT-4o
+openai_llm = create_llm("openai", model="gpt-4o")
+openai_analysis = openai_llm.generate(prompt, media=image_files)
+
+# Anthropic Claude
+claude_llm = create_llm("anthropic", model="claude-3.5-sonnet")
+claude_analysis = claude_llm.generate(prompt, media=image_files)
+
+# Local model (free)
+local_llm = create_llm("ollama", model="qwen3-vl:8b")
+local_analysis = local_llm.generate(prompt, media=image_files)
+```
+
+### 3. Document Processing Pipeline
+
+```python
+# Universal document analysis
+documents = ["contract.pdf", "financial_data.xlsx", "presentation.ppt"]
+analysis_prompt = "Extract key information and identify potential risks"
+
+# Works with any provider
+llm = create_llm("anthropic", model="claude-3.5-sonnet")
+response = llm.generate(analysis_prompt, media=documents)
+
+# Automatic format handling:
+# - PDF: Advanced text extraction with PyMuPDF4LLM
+# - Excel: Table parsing with pandas
+# - PowerPoint: Slide content extraction with unstructured
+```
+
+### 4. Local Development, Cloud Production
 
 ```python
 # Development (free, local)
@@ -411,7 +581,7 @@ llm_dev = create_llm("ollama", model="qwen3:4b-instruct-2507-q4_K_M")
 llm_prod = create_llm("openai", model="gpt-4o-mini")
 ```
 
-### 3. Embeddings & RAG
+### 5. Embeddings & RAG
 
 ```python
 from abstractcore.embeddings import EmbeddingManager
@@ -431,7 +601,7 @@ similarity = embedder.compute_similarity(query, docs[0])
 
 [Learn more about Embeddings](docs/embeddings.md)
 
-### 4. Structured Output
+### 6. Structured Output
 
 ```python
 from pydantic import BaseModel
@@ -449,7 +619,7 @@ review = llm.generate(
 print(f"{review.title}: {review.rating}/5")
 ```
 
-### 5. Universal API Server
+### 7. Universal API Server
 
 ```bash
 # Start server once
@@ -466,14 +636,16 @@ curl -X POST http://localhost:8000/v1/chat/completions \
 
 ## Why AbstractCore?
 
-- **Unified Interface**: One API for all LLM providers  
-- **Production Ready**: Robust error handling, retries, timeouts  
-- **Type Safe**: Full Pydantic integration for structured outputs  
-- **Local & Cloud**: Run models locally or use cloud APIs  
-- **Tool Calling**: Consistent function calling across providers  
-- **Streaming**: Real-time responses for interactive applications  
-- **Embeddings**: Built-in vector embeddings for RAG  
-- **Server Mode**: Optional OpenAI-compatible API server  
+- **Unified Interface**: One API for all LLM providers
+- **Multimodal Support**: Upload images, PDFs, and documents across all providers
+- **Vision Models**: Seamless integration with GPT-4o, Claude Vision, qwen3-vl, and more
+- **Production Ready**: Robust error handling, retries, timeouts
+- **Type Safe**: Full Pydantic integration for structured outputs
+- **Local & Cloud**: Run models locally or use cloud APIs
+- **Tool Calling**: Consistent function calling across providers
+- **Streaming**: Real-time responses for interactive applications
+- **Embeddings**: Built-in vector embeddings for RAG
+- **Server Mode**: Optional OpenAI-compatible API server
 - **Well Documented**: Comprehensive guides and examples  
 
 ## Installation Options
@@ -482,6 +654,9 @@ curl -X POST http://localhost:8000/v1/chat/completions \
 # Minimal core
 pip install abstractcore
 
+# With media handling (images, PDFs, documents)
+pip install abstractcore[media]
+
 # With specific providers
 pip install abstractcore[openai]
 pip install abstractcore[anthropic]
@@ -493,10 +668,25 @@ pip install abstractcore[server]
 # With embeddings
 pip install abstractcore[embeddings]
 
-# Everything
+# Everything (recommended)
 pip install abstractcore[all]
 ```
 
+**Media processing extras:**
+```bash
+# For advanced PDF processing
+pip install pymupdf4llm
+
+# For Office documents (DOCX, XLSX, PPT)
+pip install unstructured
+
+# For image optimization
+pip install pillow
+
+# For data processing (CSV, Excel)
+pip install pandas
+```
+
 ## Testing Status
 
 All tests passing as of October 12th, 2025.
@@ -519,6 +709,10 @@ All tests passing as of October 12th, 2025.
 - **[🐛 Issues](https://github.com/lpalbou/AbstractCore/issues)** - Report bugs
 - **[💬 Discussions](https://github.com/lpalbou/AbstractCore/discussions)** - Get help
 
+## Contact
+**Maintainer:** Laurent-Philippe Albou  
+📧 Email: [contact@abstractcore.ai](mailto:contact@abstractcore.ai)
+
 ## Contributing
 
 We welcome contributions! See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.

{abstractcore-2.4.2 → abstractcore-2.4.4}/README.md

@@ -1,5 +1,10 @@
 # AbstractCore
 
+[![PyPI version](https://img.shields.io/pypi/v/abstractcore.svg)](https://pypi.org/project/abstractcore/)
+[![Python Version](https://img.shields.io/pypi/pyversions/abstractcore)](https://pypi.org/project/abstractcore/)
+[![license](https://img.shields.io/github/license/lpalbou/abstractcore)](https://github.com/lpalbou/abstractcore/blob/main/LICENSE)
+[![GitHub stars](https://img.shields.io/github/stars/lpalbou/abstractcore?style=social)](https://github.com/lpalbou/abstractcore/stargazers)
+
 A unified Python library for interaction with multiple Large Language Model (LLM) providers.
 
 **Write once, run everywhere.**
@@ -64,15 +69,68 @@ loaded_session = BasicSession.load('conversation.json', provider=llm)
 
 [Learn more about Session](docs/session.md)
 
+### Media Handling
+
+AbstractCore provides **unified media handling** across all providers with **automatic maximum resolution optimization** for best results. Upload images, PDFs, and documents using the same simple API regardless of your provider.
+
+```python
+from abstractcore import create_llm
+
+# Vision analysis - works with any vision model
+# Images automatically processed at maximum supported resolution
+llm = create_llm("openai", model="gpt-4o")
+response = llm.generate(
+    "What's in this image?",
+    media=["photo.jpg"]  # Auto-resized to model's maximum capability
+)
+
+# Document analysis - works with any model
+llm = create_llm("anthropic", model="claude-3.5-sonnet")
+response = llm.generate(
+    "Summarize this research paper",
+    media=["research_paper.pdf"]
+)
+
+# Multiple files - mix images, PDFs, spreadsheets
+response = llm.generate(
+    "Analyze these business documents",
+    media=["report.pdf", "chart.png", "data.xlsx"]
+)
+
+# Same code works with local models
+llm = create_llm("ollama", model="qwen3-vl:8b")
+response = llm.generate(
+    "Describe this screenshot",
+    media=["screenshot.png"]  # Auto-optimized for qwen3-vl
+)
+```
+
+**Key Features:**
+- **Smart Resolution**: Automatically uses maximum resolution supported by each model
+- **Format Support**: PNG, JPEG, GIF, WEBP, BMP, TIFF images; PDF, TXT, MD, CSV, TSV, JSON documents
+- **Office Documents**: DOCX, XLSX, PPT (with `pip install abstractcore[all]`)
+- **Vision Optimization**: Model-specific image processing for best vision results
+
+**Provider compatibility:**
+- **High-resolution vision**: GPT-4o (up to 4096x4096), Claude 3.5 Sonnet (up to 1568x1568)
+- **Local models**: qwen3-vl (up to 3584x3584), gemma3:4b, llama3.2-vision
+- **All models**: Automatic text extraction for non-vision models
+
+[Learn more about Media Handling](docs/media-handling-system.md)
+
 ## Key Features
 
 - **Provider Agnostic**: Seamlessly switch between OpenAI, Anthropic, Ollama, LMStudio, MLX, HuggingFace
+- **Centralized Configuration**: Global defaults and app-specific preferences at `~/.abstractcore/config/abstractcore.json`
+- **Intelligent Media Handling**: Upload images, PDFs, and documents with automatic maximum resolution optimization
+- **Vision Model Support**: Smart image processing at each model's maximum capability for best results
+- **Document Processing**: Advanced PDF extraction (PyMuPDF4LLM), Office documents (DOCX/XLSX/PPT), CSV/TSV analysis
 - **Unified Tools**: Consistent tool calling across all providers
 - **Session Management**: Persistent conversations with metadata, analytics, and complete serialization
 - **Structured Responses**: Clean, predictable output formats with Pydantic
 - **Streaming Support**: Real-time token generation for interactive experiences
 - **Embeddings**: Built-in support for semantic search and RAG applications
-- **Universal Server**: Optional OpenAI-compatible API server
+- **Universal Server**: Optional OpenAI-compatible API server with `/v1/responses` endpoint
 
 ## Supported Providers
 
@@ -110,12 +168,15 @@ response = client.chat.completions.create(
 ```
 
 **Server Features:**
-- OpenAI-compatible REST endpoints (`/v1/chat/completions`, `/v1/embeddings`, etc.)
+- OpenAI-compatible REST endpoints (`/v1/chat/completions`, `/v1/embeddings`, `/v1/responses`)
+- **NEW in v2.5.0**: OpenAI Responses API (`/v1/responses`) with native `input_file` support
 - Multi-provider support through one HTTP API
+- Comprehensive media processing (images, PDFs, Office documents, CSV/TSV)
 - Agentic CLI integration (Codex, Crush, Gemini CLI)
-- Streaming responses
+- Streaming responses with optional opt-in
 - Tool call format conversion
-- Interactive API docs at `/docs`
+- Enhanced debug logging with `--debug` flag
+- Interactive API docs at `/docs` (Swagger UI)
 
 **When to use the server:**
 - Integrating with existing OpenAI-compatible tools
@@ -276,6 +337,72 @@ Each application has comprehensive documentation with examples and advanced usag
 - Integration with shell scripts and automation
 - Standardized text processing tasks
 
+## Configuration
+
+AbstractCore provides a **centralized configuration system** that manages default models, cache directories, and logging settings from a single location. This eliminates the need to specify `--provider` and `--model` parameters repeatedly.
+
+### Quick Setup
+
+```bash
+# Check current configuration (shows how to change each setting)
+abstractcore --status
+
+# Set defaults for all applications
+abstractcore --set-global-default ollama/llama3:8b
+
+# Or configure specific applications (examples of customization)
+abstractcore --set-app-default summarizer openai gpt-4o-mini
+abstractcore --set-app-default extractor ollama qwen3:4b-instruct
+abstractcore --set-app-default judge anthropic claude-3-5-haiku
+
+# Configure logging (common examples)
+abstractcore --set-console-log-level WARNING  # Reduce console output
+abstractcore --set-console-log-level NONE     # Disable console logging
+abstractcore --enable-file-logging            # Save logs to files
+abstractcore --enable-debug-logging           # Full debug mode
+
+# Set API keys as needed
+abstractcore --set-api-key openai sk-your-key-here
+abstractcore --set-api-key anthropic your-anthropic-key
+
+# Verify configuration (includes change commands for each setting)
+abstractcore --status
+```
+
+### Priority System
+
+AbstractCore uses a clear priority system where explicit parameters always override defaults:
+
+1. **Explicit parameters** (highest priority): `summarizer doc.txt --provider openai --model gpt-4o-mini`
+2. **App-specific config**: `abstractcore --set-app-default summarizer openai gpt-4o-mini`
+3. **Global config**: `abstractcore --set-global-default openai/gpt-4o-mini`
+4. **Built-in defaults** (lowest priority): `huggingface/unsloth/Qwen3-4B-Instruct-2507-GGUF`
+
+### Usage After Configuration
+
+Once configured, apps use your defaults automatically:
+
+```bash
+# Before configuration (requires explicit parameters)
+summarizer document.pdf --provider openai --model gpt-4o-mini
+
+# After configuration (uses configured defaults)
+summarizer document.pdf
+
+# Explicit parameters still override when needed
+summarizer document.pdf --provider anthropic --model claude-3-5-sonnet
+```
+
+### Configuration Features
+
+- **Application defaults**: Different optimal models for each app
+- **Cache directories**: Configurable cache locations for models and data
+- **Logging control**: Package-wide logging levels and debug mode
+- **API key management**: Centralized API key storage
+- **Interactive setup**: `abstractcore --configure` for guided configuration
+
+**Complete guide**: [Centralized Configuration](docs/centralized-config.md)
+
 ## Documentation
 
 **📚 Complete Documentation:** [docs/](docs/) - Full documentation index and navigation guide
@@ -287,6 +414,7 @@ Each application has comprehensive documentation with examples and advanced usag
 
 ### Core Library (Python)
 - **[Python API Reference](docs/api-reference.md)** - Complete Python API documentation
+- **[Media Handling System](docs/media-handling-system.md)** - Images, PDFs, and document processing across all providers
 - **[Session Management](docs/session.md)** - Persistent conversations, serialization, and analytics
 - **[Embeddings Guide](docs/embeddings.md)** - Semantic search, RAG, and vector embeddings
 - **[Code Examples](examples/)** - Working examples for all features
@@ -312,7 +440,44 @@ for provider in providers:
     response = llm.generate("Hello!")
 ```
 
-### 2. Local Development, Cloud Production
+### 2. Vision Analysis Across Providers
+
+```python
+# Same image analysis works with any vision model
+image_files = ["product_photo.jpg", "user_feedback.png"]
+prompt = "Analyze these product images and suggest improvements"
+
+# OpenAI GPT-4o
+openai_llm = create_llm("openai", model="gpt-4o")
+openai_analysis = openai_llm.generate(prompt, media=image_files)
+
+# Anthropic Claude
+claude_llm = create_llm("anthropic", model="claude-3.5-sonnet")
+claude_analysis = claude_llm.generate(prompt, media=image_files)
+
+# Local model (free)
+local_llm = create_llm("ollama", model="qwen3-vl:8b")
+local_analysis = local_llm.generate(prompt, media=image_files)
+```
+
+### 3. Document Processing Pipeline
+
+```python
+# Universal document analysis
+documents = ["contract.pdf", "financial_data.xlsx", "presentation.ppt"]
+analysis_prompt = "Extract key information and identify potential risks"
+
+# Works with any provider
+llm = create_llm("anthropic", model="claude-3.5-sonnet")
+response = llm.generate(analysis_prompt, media=documents)
+
+# Automatic format handling:
+# - PDF: Advanced text extraction with PyMuPDF4LLM
+# - Excel: Table parsing with pandas
+# - PowerPoint: Slide content extraction with unstructured
+```
+
+### 4. Local Development, Cloud Production
 
 ```python
 # Development (free, local)
@@ -322,7 +487,7 @@ llm_dev = create_llm("ollama", model="qwen3:4b-instruct-2507-q4_K_M")
 llm_prod = create_llm("openai", model="gpt-4o-mini")
 ```
 
-### 3. Embeddings & RAG
+### 5. Embeddings & RAG
 
 ```python
 from abstractcore.embeddings import EmbeddingManager
@@ -342,7 +507,7 @@ similarity = embedder.compute_similarity(query, docs[0])
 
 [Learn more about Embeddings](docs/embeddings.md)
 
-### 4. Structured Output
+### 6. Structured Output
 
 ```python
 from pydantic import BaseModel
@@ -360,7 +525,7 @@ review = llm.generate(
 print(f"{review.title}: {review.rating}/5")
 ```
 
-### 5. Universal API Server
+### 7. Universal API Server
 
 ```bash
 # Start server once
@@ -377,14 +542,16 @@ curl -X POST http://localhost:8000/v1/chat/completions \
 
 ## Why AbstractCore?
 
-- **Unified Interface**: One API for all LLM providers  
-- **Production Ready**: Robust error handling, retries, timeouts  
-- **Type Safe**: Full Pydantic integration for structured outputs  
-- **Local & Cloud**: Run models locally or use cloud APIs  
-- **Tool Calling**: Consistent function calling across providers  
-- **Streaming**: Real-time responses for interactive applications  
-- **Embeddings**: Built-in vector embeddings for RAG  
-- **Server Mode**: Optional OpenAI-compatible API server  
+- **Unified Interface**: One API for all LLM providers
+- **Multimodal Support**: Upload images, PDFs, and documents across all providers
+- **Vision Models**: Seamless integration with GPT-4o, Claude Vision, qwen3-vl, and more
+- **Production Ready**: Robust error handling, retries, timeouts
+- **Type Safe**: Full Pydantic integration for structured outputs
+- **Local & Cloud**: Run models locally or use cloud APIs
+- **Tool Calling**: Consistent function calling across providers
+- **Streaming**: Real-time responses for interactive applications
+- **Embeddings**: Built-in vector embeddings for RAG
+- **Server Mode**: Optional OpenAI-compatible API server
 - **Well Documented**: Comprehensive guides and examples  
 
 ## Installation Options
@@ -393,6 +560,9 @@ curl -X POST http://localhost:8000/v1/chat/completions \
 # Minimal core
 pip install abstractcore
 
+# With media handling (images, PDFs, documents)
+pip install abstractcore[media]
+
 # With specific providers
 pip install abstractcore[openai]
 pip install abstractcore[anthropic]
@@ -404,10 +574,25 @@ pip install abstractcore[server]
 # With embeddings
 pip install abstractcore[embeddings]
 
-# Everything
+# Everything (recommended)
 pip install abstractcore[all]
 ```
 
+**Media processing extras:**
+```bash
+# For advanced PDF processing
+pip install pymupdf4llm
+
+# For Office documents (DOCX, XLSX, PPT)
+pip install unstructured
+
+# For image optimization
+pip install pillow
+
+# For data processing (CSV, Excel)
+pip install pandas
+```
+
 ## Testing Status
 
 All tests passing as of October 12th, 2025.
@@ -430,6 +615,10 @@ All tests passing as of October 12th, 2025.
 - **[🐛 Issues](https://github.com/lpalbou/AbstractCore/issues)** - Report bugs
 - **[💬 Discussions](https://github.com/lpalbou/AbstractCore/discussions)** - Get help
 
+## Contact
+**Maintainer:** Laurent-Philippe Albou  
+📧 Email: [contact@abstractcore.ai](mailto:contact@abstractcore.ai)
+
 ## Contributing
 
 We welcome contributions! See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
@@ -444,4 +633,4 @@ MIT License - see [LICENSE](LICENSE) file for details.
 
 ---
 
-> **Migration Note**: This project was previously known as "AbstractLLM" and has been completely rebranded to "AbstractCore" as of version 2.4.0. See [CHANGELOG.md](CHANGELOG.md) for migration details.
+> **Migration Note**: This project was previously known as "AbstractLLM" and has been completely rebranded to "AbstractCore" as of version 2.4.0. See [CHANGELOG.md](CHANGELOG.md) for migration details.