### Install Cactus CLI from Source (macOS)

Source: https://cactuscompute.com/docs/v1.7/cli

Installs the Cactus CLI by cloning the repository and running the setup script. This method is suitable for macOS users who want to build from source.

```bash
git clone https://github.com/cactus-compute/cactus && cd cactus && source ./setup
```

--------------------------------

### Install Cactus CLI with Homebrew

Source: https://cactuscompute.com/docs/v1.7/cli

Installs the Cactus CLI using the Homebrew package manager on macOS. This is the recommended and simplest way to get started.

```bash
brew install cactus-compute/cactus/cactus
```

--------------------------------

### Install Cactus CLI from Source (Linux)

Source: https://cactuscompute.com/docs/v1.7/cli

Installs the Cactus CLI from source on Linux systems. This requires installing prerequisite packages like python3, pip, cmake, and build essentials.

```bash
sudo apt-get install python3 python3-venv python3-pip cmake build-essential libcurl4-openssl-dev
git clone https://github.com/cactus-compute/cactus && cd cactus && source ./setup
```

--------------------------------

### Run a Local AI Model

Source: https://cactuscompute.com/docs/v1.7/cli

Starts an interactive playground to run a specified AI model locally. The CLI automatically downloads the model if it's not already present.

```bash
# Download and run LiquidAI's LFM2-350M
cactus run LiquidAI/LFM2-350M

# Or use a specific model
cactus run google/gemma-3-270m-it
```

--------------------------------

### Include Cactus Header in C++

Source: https://cactuscompute.com/docs/v1.7/cpp

This snippet shows how to include the main Cactus header file in your C++ project to start using the SDK. Ensure the Cactus library is correctly linked during the build process.

```cpp
#include <cactus.h>
```

--------------------------------

### Download AI Models

Source: https://cactuscompute.com/docs/v1.7/cli

Downloads AI model weights for offline use. Models are stored in the `./weights/` directory. Supports specifying model precision.

```bash
# Download a model for offline use
cactus download LiquidAI/LFM2.5-1.2B-Instruct

# Download with specific precision
cactus download LiquidAI/LFM2.5-1.2B-Instruct --precision int8

# Models are stored in ./weights/
```

--------------------------------

### Basic Text Completion with Cactus C++ SDK

Source: https://cactuscompute.com/docs/v1.7/cpp

Demonstrates a basic text completion request using the Cactus C++ SDK. It initializes a model, prepares messages, and generates a response. Optional generation settings can be provided.

```cpp
#include <cactus.h>

// Initialize model
cactus_model_t model = cactus_init(
    "path/to/weight/folder",
    "path/to/rag/documents", // optional: for auto-RAG
);

// Prepare messages
const char* messages = R"([\n    {"role": "system", "content": "You are a helpful assistant."},\n    {"role": "user", "content": "What is the capital of France?"}\n])";

// Optional: generation settings
const char* options = R"({\n    "max_tokens": 50,\n    "stop_sequences": ["<|im_end|>"]\n})";

// Generate response
char response[4096];
int result = cactus_complete(
    model,           // model handle
    messages,        // chat messages (JSON array)
    response,        // output buffer
    sizeof(response),// buffer size
    options,         // generation options (or nullptr)
    nullptr,         // tools for function calling (or nullptr)
    nullptr,         // streaming callback (or nullptr)
    nullptr          // user data for callback
);

```

--------------------------------

### Building Computation Graph with Cactus Graph API (C++)

Source: https://cactuscompute.com/docs/v1.7/cpp

Demonstrates how to define and execute a custom computation graph using the Cactus Graph API in C++. This allows for the implementation of custom models with optimized performance.

```cpp
#include <cactus.h>

CactusGraph graph;

// Define inputs
auto a = graph.input({2, 3}, Precision::FP16);
auto b = graph.input({3, 4}, Precision::INT8);

// Build computation graph
auto x1 = graph.matmul(a, b, false);
auto x2 = graph.transpose(x1);
auto result = graph.matmul(b, x2, true);

// Set input data
float a_data[6] = {1.1f, 2.3f, 3.4f, 4.2f, 5.7f, 6.8f};
float b_data[12] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12};

graph.set_input(a, a_data, Precision::FP16);
graph.set_input(b, b_data, Precision::INT8);

// Execute
graph.execute();

// Get output
void* output_data = graph.get_output(result);

// Clean up
graph.hard_reset();

```

--------------------------------

### Streaming Responses with Cactus C++ SDK

Source: https://cactuscompute.com/docs/v1.7/cpp

Shows how to implement streaming responses using a callback function with the Cactus C++ SDK. The `token_callback` function is invoked for each generated token, allowing real-time output.

```cpp
void token_callback(const char* token, int token_id, void* user_data) {
    printf("%s", token);
    fflush(stdout);
}

cactus_complete(
    model,
    messages,
    response,
    sizeof(response),
    nullptr,
    nullptr,
    token_callback,  // streaming callback
    nullptr          // user data
);

```

--------------------------------

### Function Calling with Cactus C++ SDK

Source: https://cactuscompute.com/docs/v1.7/cpp

Illustrates how to enable function calling with the Cactus C++ SDK. By providing a 'tools' JSON array, the SDK can identify and format function calls based on user input.

```cpp
const char* tools = R"([\n    {\n        "name": "get_weather",\n        "description": "Get current weather for a location",\n        "parameters": {\n            "type": "object",\n            "properties": {\n                "location": {"type": "string"}\n            },\n            "required": ["location"]\n        }\n    }\n])";

char response[4096];
cactus_complete(
    model,
    messages,
    response,
    sizeof(response),
    nullptr,  // use default options
    tools,    // pass tools JSON
    nullptr,
    nullptr
);

```

--------------------------------

### Configure Cloud Authentication

Source: https://cactuscompute.com/docs/v1.7/cli

Sets up cloud API credentials for Cactus, enabling hybrid cloud routing. This allows the CLI to automatically switch to cloud-based models when local performance is insufficient.

```bash
cactus auth
```

--------------------------------

### Perform Audio Transcription

Source: https://cactuscompute.com/docs/v1.7/cli

Transcribes audio from a file or live microphone input using a specified model. Supports cloud fallback for improved accuracy on noisy audio.

```bash
# Transcribe an audio file
cactus transcribe openai/whisper-small --file recording.mp3

# Live microphone transcription
cactus transcribe UsefulSensors/moonshine-base

# With cloud fallback for noisy audio
cactus transcribe openai/whisper-small --file recording.mp3 --cloud-key YOUR_API_KEY
```

--------------------------------

### Test AI Models on Different Platforms

Source: https://cactuscompute.com/docs/v1.7/cli

Runs tests for AI models on specified platforms like iOS or Android. Allows for targeting specific models and precision levels during testing.

```bash
# Test on iOS simulator
cactus test --ios --model LiquidAI/LFM2-350M

# Test on Android with specific precision
cactus test --android --model google/gemma-3-270m-it --precision int8
```

--------------------------------

### Convert Model with LoRA Adapter

Source: https://cactuscompute.com/docs/v1.7/cli

Converts a model to the `.cact` format and supports merging LoRA adapters during the conversion process. This is useful for applying fine-tuning to existing models.

```bash
# Convert a model with LoRA adapter
cactus convert LiquidAI/LFM2-350M ./output --lora path/to/lora
```

--------------------------------

### Handle Cactus Compute API Errors (C++)

Source: https://cactuscompute.com/docs/v1.7/cpp

This snippet demonstrates how to check for and handle errors returned by the Cactus Compute API. It assumes the API returns an integer status code, where non-zero indicates an error. Error details can be parsed from a JSON response.

```c++
int result = cactus_complete(...);

if (result != 0) {
    // Parse error from response JSON
    // error field will contain specific error message
}
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.