API Reference on Llamatik Documentation

Models Status

Mon, 01 Jan 0001 00:00:00 +0000

Models Status

`GET /v1/models`

Returns which models are currently loaded in each inference bridge, plus the capabilities of the generation model. Use it to discover server state (for example, after startup loading) without calling each bridge separately.

Only model filenames are returned — full server paths are never exposed.

curl -s http://localhost:8080/v1/models | jq

{
 "generation": {
 "loaded": true,
 "model": "llama-3.2-3b-instruct-q8_0.gguf",
 "capabilities": { "finetuneType": "instruct", "hasChatTemplate": true }
 },
 "embedding": { "loaded": false, "model": null },
 "whisper": { "loaded": true, "model": "ggml-base-q8_0.bin" },
 "stableDiffusion": {
 "loaded": false, "model": null,
 "vae": null, "t5xxl": null, "clipVision": null, "supportsVideo": null
 },
 "multimodal": { "loaded": false, "model": null, "mmproj": null },
 "mtp": { "loaded": false, "model": null }
}

Field	Description
`generation.capabilities`	Present only when a generation model is loaded. `finetuneType` is the GGUF `general.finetune` value (`instruct`, `chat`, `base` or `null`). `hasChatTemplate` is `true` when `POST /v1/generation/model/applyChatTemplate` can render messages for this model.
`stableDiffusion.supportsVideo`	`true` when the loaded model supports `txt2video`; `null` when nothing is loaded.
`stableDiffusion.vae` / `t5xxl` / `clipVision`	Filenames of companion files, when loaded.

The state is updated by the /init, /shutdown and /release endpoints as well as by startup loading.

Generation

Mon, 01 Jan 0001 00:00:00 +0000

Generation — `/v1/generation`

The generation endpoints wrap LlamaBridge text generation, streaming, JSON output, KV session management, speculative decoding (MTP), and model introspection.

Initialize

`POST /v1/generation/init`

Load a generation model. Must be called before any other generation endpoint.

Request:

{ "modelPath": "/models/tinyllama.gguf" }

Response:

{ "ok": true }

Returns 400 if the model fails to load (message sanitized — no filesystem paths in the response).

Text generation

`POST /v1/generation/generate`

One-shot generation — returns when the full output is ready.

Embeddings

Mon, 01 Jan 0001 00:00:00 +0000

Embeddings — `/v1/embeddings`

The embeddings endpoints wrap LlamaBridge embedding inference. Use this for RAG pipelines, semantic search, and similarity ranking.

Initialize

`POST /v1/embeddings/init`

Load an embedding model. Must be called before /embed.

Request:

{ "modelPath": "/models/nomic-embed-text-v1.5.Q8_0.gguf" }

Response:

{ "ok": true }

Returns 400 on failure.

Generate an embedding

`POST /v1/embeddings/embed`

Compute an embedding vector for the given input string.

Request:

{ "input": "Kotlin Multiplatform" }

Response:

{ "embedding": [0.12, -0.04, 0.87, "..."] }

The embedding array length matches the model’s output dimension. Typical values:

Whisper

Mon, 01 Jan 0001 00:00:00 +0000

Whisper — `/v1/whisper`

The Whisper endpoints wrap WhisperBridge speech-to-text. Two transcription modes are available: flat text (transcribe) and structured per-segment JSON (transcribeSegments).

Initialize

`POST /v1/whisper/init`

Load a Whisper GGML model. Must be called before any transcription endpoint.

Request:

{ "modelPath": "/models/ggml-base.en.bin" }

Response:

{ "ok": true }

Transcribe

`POST /v1/whisper/transcribe`

Transcribe a WAV file to a flat string.

Request:

{
 "wavPath": "/audio/recording.wav",
 "language": "en",
 "initialPrompt": "Technical vocabulary"
}

language and initialPrompt are optional.

language — ISO 639-1 code (e.g. "en", "de", "fr"). Omit to auto-detect.
initialPrompt — prime the model with vocabulary or context to improve accuracy on domain-specific content.

Response:

Stable Diffusion

Mon, 01 Jan 0001 00:00:00 +0000

Stable Diffusion — `/v1/stablediffusion`

The Stable Diffusion endpoints wrap StableDiffusionBridge for text-to-image, image-to-image, and text-to-video generation.

Images are exchanged as Base64-encoded RGBA bytes (width × height × 4 bytes). Video frames use a packed binary format described below.

Initialize

`POST /v1/stablediffusion/init`

Load a model. For standard image models only modelPath is required. Video models (Wan 2.1/2.2) additionally require vaePath and typically t5xxlPath.

Standard image model:

{
 "modelPath": "/models/dreamshaper.safetensors",
 "threads": 4
}

Video model (Wan 2.1 text-to-video):

Multimodal

Mon, 01 Jan 0001 00:00:00 +0000

Multimodal — `/v1/multimodal`

The multimodal endpoints wrap MultimodalBridge for vision-language model (VLM) inference. Image analysis is returned as a streaming SSE response.

Initialize

`POST /v1/multimodal/init`

Load a VLM and its mmproj companion. Must be called before analyzeStream.

Request:

{
 "modelPath": "/models/SmolVLM-256M-Instruct-Q8_0.gguf",
 "mmprojPath": "/models/mmproj-SmolVLM-256M-f16.gguf"
}

Response:

{ "ok": true }

Both paths must resolve inside MODELS_DIR if that variable is configured.

Analyze image (streaming)

`POST /v1/multimodal/analyzeStream`

Analyze an image with a text prompt. Returns Content-Type: text/event-stream.

API Reference on Llamatik Documentation

Models Status

Models Status

GET /v1/models

Generation

Generation — /v1/generation

Initialize

POST /v1/generation/init

Text generation

POST /v1/generation/generate

Embeddings

Embeddings — /v1/embeddings

Initialize

POST /v1/embeddings/init

Generate an embedding

POST /v1/embeddings/embed

Whisper

Whisper — /v1/whisper

Initialize

POST /v1/whisper/init

Transcribe

POST /v1/whisper/transcribe

Stable Diffusion

Stable Diffusion — /v1/stablediffusion

Initialize

POST /v1/stablediffusion/init

Multimodal

Multimodal — /v1/multimodal

Initialize

POST /v1/multimodal/init

Analyze image (streaming)

POST /v1/multimodal/analyzeStream

`GET /v1/models`

Generation — `/v1/generation`

`POST /v1/generation/init`

`POST /v1/generation/generate`

Embeddings — `/v1/embeddings`

`POST /v1/embeddings/init`

`POST /v1/embeddings/embed`

Whisper — `/v1/whisper`

`POST /v1/whisper/init`

`POST /v1/whisper/transcribe`

Stable Diffusion — `/v1/stablediffusion`

`POST /v1/stablediffusion/init`

Multimodal — `/v1/multimodal`

`POST /v1/multimodal/init`

`POST /v1/multimodal/analyzeStream`