Struct MtmdCapabilities 

pub struct MtmdCapabilities {
    pub chat: bool,
    pub vision: bool,
    pub audio: bool,
    pub video: bool,
    pub audio_stream: bool,
    pub text_gen: bool,
    pub audio_gen: bool,
    pub audio_gen_spk_ref: bool,
}

What a model and projector pair can take in and produce.

From mtmd_helper_get_capabilities. The three input flags answer “what can I feed this?”; the three output flags answer “what will it give me back?” — a text-to-speech model, for example, reports text_gen: false and audio_gen: true, which no earlier predicate could express.

Fields

chat: bool

Usable for chat. false for tts-only models (no chat template) and for realtime transcription models.

vision: bool

Accepts image input.

audio: bool

Accepts audio input.

video: bool

Accepts video input. Also depends on the MTMD_VIDEO build option, and on ffmpeg/ffprobe at runtime — see MtmdVideo.

audio_stream: bool

Supports realtime transcription through llama.cpp’s mtmd_audio_stream_* API, which this crate does not wrap yet — reach for llama-cpp-sys-4 directly if you need it.

text_gen: bool

Generates text. false for an audio-only model.

audio_gen: bool

Generates audio.

audio_gen_spk_ref: bool

Accepts a speaker reference clip for voice cloning.

Trait Implementations

impl Clone for MtmdCapabilities

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for MtmdCapabilities

impl Debug for MtmdCapabilities

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for MtmdCapabilities

impl PartialEq for MtmdCapabilities

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for MtmdCapabilities