Struct ModelSummary 

pub struct ModelSummary {

    pub id: u32,
    pub name: String,
    pub tags: Vec<String>,
    pub capabilities: ModelCapabilities,
    pub param_count: String,
    pub quantization: Option<String>,
    pub architecture: Option<String>,
    pub created_at: i64,
    pub file_size: u64,
    pub context_length: Option<u64>,
    pub inference_defaults: Option<InferenceConfig>,
    pub defaults_origin: Option<DefaultsOrigin>,
    pub server_defaults: Option<ServerConfig>,
}

Domain model summary for catalog operations (listing).

This is a domain type (not an OpenAI API type). The proxy layer is responsible for mapping this to OpenAI-compatible formats.

Note: Does NOT include file_path to avoid leaking filesystem details in catalog/listing operations.

Fields

id: u32

Database ID of the model.

name: String

Model name (used as identifier).

tags: Vec<String>

Tags/labels associated with the model.

capabilities: ModelCapabilities

Detected and persisted capability flags for this model.

This is the single source of truth for model behaviour constraints (strict-turn alternation, system-role support, tool calls, reasoning). The proxy uses these directly rather than inferring from tags at request time, eliminating the split-brain between tags and capabilities.

param_count: String

Parameter count as string (e.g., “7B”, “13B”, “70B”).

quantization: Option<String>

Quantization type (e.g., “Q4_K_M”, “Q8_0”).

architecture: Option<String>

Model architecture (e.g., “llama”, “mistral”, “qwen2”).

created_at: i64

Unix timestamp when the model was added.

file_size: u64

File size in bytes.

context_length: Option<u64>

Maximum context length the model supports, in tokens (from GGUF metadata). None when unknown.

This is a static, per-model ceiling — it does not reflect the --ctx-size a currently-running instance was actually launched with, which can be smaller (see gglib_core::ports::model_runtime::RunningTarget::effective_ctx for the live value). Consumers that need the true, currently-running context size should prefer effective_ctx when the model is running and fall back to this field otherwise.

inference_defaults: Option<InferenceConfig>

Per-model inference parameter defaults.

When Some, these are resolved per-request via InferenceConfig::resolve_with_defaults before forwarding to llama-server. Used by gglib proxy to inject resolved defaults into OpenAI-format request bodies, and by the agentic loop (gglib chat, gglib q) to apply model-specific sampling parameters.

defaults_origin: Option<DefaultsOrigin>

Whether Self::inference_defaults was set by the user or auto-detected. See DefaultsOrigin and InferenceConfig::resolve_with_profile.

server_defaults: Option<ServerConfig>

Per-model server defaults (context_length, etc.) from the database.

Implementations

impl ModelSummary

pub fn description(&self) -> String

Create a description string for this model.

Trait Implementations

impl Clone for ModelSummary

fn clone(&self) -> ModelSummary

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModelSummary

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

Formats the value using the given formatter.

impl From<&ModelSummary> for ModelContext

fn from(summary: &ModelSummary) -> Self

Converts to this type from the input type.