> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nexa.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Basics
> API structures and configuration options for Nexa Android SDK.
## **Plugin Selection**
Choose the appropriate backend plugin for your model type and format.
```kotlin theme={"dark"}
val plugin_id: String? // Backend plugin to use
```
**Available Plugins:**
* `"cpu_gpu"` - GGUF backend for CPU/GPU/Hexagon NPU (LLM, VLM). Device is selected via `device_id` + `nGpuLayers`.
* `"npu"` - NPU backend for NEXA format models (LLM, VLM, Embeddings, ASR, CV, Rerank)
* `"whisper_cpp"` - Whisper.cpp backend for ASR
* `"tts_cpp"` - TTS backend for text-to-speech
***
## **Device Selection**
Control which hardware device processes your model.
```kotlin theme={"dark"}
val device_id: String? // Device to use for the model
```
**Available Values:**
* `null` - CPU (default)
* `"gpu"` - GPU acceleration via OpenCL
* `"dev0"` - Qualcomm Hexagon NPU acceleration
**Hardware Acceleration with GGUF (`plugin_id = "cpu_gpu"`):**
* **GPU**: set `device_id = "gpu"` and set `nGpuLayers > 0` in `ModelConfig`
* **Hexagon NPU (GGML backend)**: set `device_id = "dev0"` and set `nGpuLayers > 0` in `ModelConfig`
If `nGpuLayers = 0` (or `device_id = null`), the model runs on CPU.
***
## **LLM Data Structures**
### LlmCreateInput
```kotlin theme={"dark"}
data class LlmCreateInput(
val model_name: String? = null, // Model name (required for NPU, empty for CPU/GPU)
val model_path: String, // Path to model file or folder
val tokenizer_path: String? = null, // Optional tokenizer path
val config: ModelConfig, // Model configuration
val plugin_id: String? = null, // "npu", "cpu_gpu", etc.
val device_id: String? = null // null, "gpu", "dev0"
)
```
### ChatMessage
```kotlin theme={"dark"}
data class ChatMessage(
val role: String, // "system", "user", "assistant"
val content: String // Message text
)
```
### GenerationConfig
```kotlin theme={"dark"}
data class GenerationConfig(
val maxTokens: Int? = null, // Maximum tokens to generate
val temperature: Float = 0.7f, // Sampling temperature
val topP: Float = 0.9f, // Nucleus sampling
val topK: Int = 40, // Top-K sampling
val repeatPenalty: Float = 1.1f, // Repetition penalty
val seed: Int = -1 // Random seed (-1 for random)
)
```
### LlmStreamResult
```kotlin theme={"dark"}
sealed class LlmStreamResult {
data class Token(val text: String) : LlmStreamResult()
object Completed : LlmStreamResult()
data class Error(val throwable: Throwable) : LlmStreamResult()
}
```
***
## **Multimodal Data Structures**
### VlmCreateInput
```kotlin theme={"dark"}
data class VlmCreateInput(
val model_name: String? = null, // Model name (required for NPU)
val model_path: String, // Path to VLM model
val tokenizer_path: String? = null, // Optional tokenizer path
val mmproj_path: String? = null, // Vision projection weights (for GGUF models)
val config: ModelConfig, // Model configuration
val plugin_id: String? = null, // "npu" for NPU, "cpu_gpu" for CPU/GPU
val device_id: String? = null // null for CPU, "gpu" for GPU, "dev0" for NPU
)
```
### VlmChatMessage
```kotlin theme={"dark"}
data class VlmChatMessage(
val role: String, // "system", "user", "assistant"
val contents: List // Mixed text/image/audio content
)
```
### VlmContent
```kotlin theme={"dark"}
data class VlmContent(
val type: String, // "text", "image", "audio"
val content: String // Text or file path
)
```
***
## **Embeddings Data Structures**
### EmbedderCreateInput
```kotlin theme={"dark"}
data class EmbedderCreateInput(
val model_name: String? = null, // Model name for NPU
val model_path: String, // Path to embedder model
val tokenizer_path: String? = null, // Path to tokenizer
val config: ModelConfig, // Model configuration
val plugin_id: String? = null, // "npu" for NPU, null for CPU
val device_id: String? = null // Device ID (optional)
)
```
### EmbeddingConfig
```kotlin theme={"dark"}
data class EmbeddingConfig(
val normalize: Boolean = true // Normalize embeddings (default: true)
)
```
***
## **ASR Data Structures**
### AsrCreateInput
```kotlin theme={"dark"}
data class AsrCreateInput(
val model_name: String? = null, // Model name for NPU
val model_path: String, // Path to ASR model
val config: ModelConfig, // Model configuration
val plugin_id: String? = null, // "whisper_cpp" for whisper.cpp, "npu" for NPU
val device_id: String? = null // Device ID (optional)
)
```
### AsrTranscribeInput
```kotlin theme={"dark"}
data class AsrTranscribeInput(
val audioPath: String, // Path to audio file (.wav, .mp3, etc.)
val language: String, // Language code: "en", "zh", "es", etc.
val timestamps: String? = null // Optional: timestamp format
)
```
### AsrTranscriptionResult
```kotlin theme={"dark"}
data class AsrTranscriptionResult(
val result: AsrResult, // Transcription result
val profileData: String // Performance metrics
)
```
***
## **TTS Data Structures**
### TtsCreateInput
```kotlin theme={"dark"}
data class TtsCreateInput(
val model_name: String? = null, // Model name (empty for CPU/GPU)
val model_path: String, // Path to TTS model
val vocoder_path: String? = null, // Optional vocoder path
val config: ModelConfig, // Model configuration
val plugin_id: String? = null, // "tts_cpp" for TTS backend
val device_id: String? = null // Device ID (optional)
)
```
### TtsSynthesizeInput
```kotlin theme={"dark"}
data class TtsSynthesizeInput(
val textUtf8: String, // Text to synthesize (UTF-8 encoding)
val config: TtsConfig? = null, // Optional TTS configuration
val outputPath: String? = null // Optional output file path (null for auto-generated)
)
```
### TtsConfig
```kotlin theme={"dark"}
data class TtsConfig(
val voice: String? = null, // Voice identifier
val speed: Float = 1.0f, // Speech speed (1.0 = normal)
val seed: Int = -1, // Random seed (-1 for random)
val sampleRate: Int = 22050 // Output sample rate in Hz
)
```
### TtsSynthesizeOutput
```kotlin theme={"dark"}
data class TtsSynthesizeOutput(
val outputPath: String, // Path to generated audio file
val profileData: String // Performance metrics
)
```
***
## **Rerank Data Structures**
### RerankerCreateInput
```kotlin theme={"dark"}
data class RerankerCreateInput(
val model_name: String? = null, // Model name for NPU
val model_path: String, // Path to reranker model
val tokenizer_path: String? = null, // Path to tokenizer
val config: ModelConfig, // Model configuration
val plugin_id: String? = null, // "npu" for NPU, null for CPU
val device_id: String? = null // Device ID (optional)
)
```
### RerankConfig
```kotlin theme={"dark"}
data class RerankConfig(
val topN: Int? = null // Return only top N results (null = all)
)
```
### RerankerResult
```kotlin theme={"dark"}
data class RerankerResult(
val scores: FloatArray, // Relevance scores (0.0 - 1.0)
val scoreCount: Int, // Number of scores
val profileData: String // Performance metrics
)
```
***
## **Computer Vision Data Structures**
### CVCreateInput
```kotlin theme={"dark"}
data class CVCreateInput(
val model_name: String, // Model name: "paddleocr", etc.
val config: CVModelConfig, // CV model configuration
val plugin_id: String? = null // "npu" for NPU, "cpu_gpu" for CPU/GPU
)
```
### CVModelConfig
```kotlin theme={"dark"}
data class CVModelConfig(
val capabilities: CVCapability, // OCR, DETECTION, CLASSIFICATION
val det_model_path: String? = null, // Detection model path
val rec_model_path: String? = null, // Recognition model path
val char_dict_path: String? = null, // Character dictionary path
val qnn_model_folder_path: String? = null, // QNN model folder (NPU)
val qnn_lib_folder_path: String? = null // QNN library path (NPU)
)
```
### CVCapability
```kotlin theme={"dark"}
enum class CVCapability {
OCR, // Optical Character Recognition
DETECTION, // Object detection
CLASSIFICATION // Image classification
}
```
### CVResult
```kotlin theme={"dark"}
data class CVResult(
val text: String, // Recognized text (OCR)
val confidence: Float, // Confidence score (0.0 - 1.0)
val boundingBox: BoundingBox?, // Bounding box coordinates
val label: String?, // Class label (classification)
val score: Float // Detection/classification score
)
```
***
## **Need Help?**
Join our community to get support, share your projects, and connect with other developers.
Get real-time support and chat with the Nexa AI community
Collaborate with developers and access community resources