LMM

Class: LMMBlockV1

Source: inference.core.workflows.core_steps.models.foundation.lmm.v1.LMMBlockV1

Ask a question to a Large Multimodal Model (LMM) with an image and text.

You can specify arbitrary text prompts to an LMMBlock.

The LLMBlock supports two LMMs:

• OpenAI's GPT-4 with Vision;

You need to provide your OpenAI API key to use the GPT-4 with Vision model.

If you want to classify an image into one or more categories, we recommend using the dedicated LMMForClassificationBlock.

Type identifier

Use the following identifier in step "type" field: roboflow_core/lmm@v1to add the block as as step in your workflow.

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌
prompt	str	Holds unconstrained text prompt to LMM mode.	✅
lmm_type	str	Type of LMM to be used.	✅
lmm_config	LMMConfig	Configuration of LMM.	❌
remote_api_key	str	Holds API key required to call LMM model - in current state of development, we require OpenAI key when lmm_type=gpt_4v..	✅
json_output	Dict[str, str]	Holds dictionary that maps name of requested output field into its description.	❌

The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.

Available Connections

Compatible Blocks

Check what blocks you can connect to LMM in version v1.

• inputs: VLM as Detector, Google Vision OCR, Classification Label Visualization, Circle Visualization, Image Contours, Relative Static Crop, Image Preprocessing, LMM For Classification, VLM as Classifier, Ellipse Visualization, Stitch Images, Triangle Visualization, Stability AI Inpainting, QR Code Generator, Image Slicer, Background Color Visualization, Model Monitoring Inference Aggregator, OCR Model, Dot Visualization, Florence-2 Model, SIFT, Morphological Transformation, EasyOCR, Reference Path Visualization, Halo Visualization, SIFT Comparison, Polygon Visualization, Image Slicer, Florence-2 Model, Slack Notification, Clip Comparison, Image Convert Grayscale, Instance Segmentation Model, OpenAI, Color Visualization, Keypoint Detection Model, Google Gemini, Label Visualization, Email Notification, Llama 3.2 Vision, Trace Visualization, Email Notification, Corner Visualization, Mask Visualization, CogVLM, Stability AI Outpainting, OpenAI, Roboflow Custom Metadata, Stitch OCR Detections, Blur Visualization, CSV Formatter, Crop Visualization, OpenAI, Grid Visualization, Perspective Correction, Twilio SMS Notification, Absolute Static Crop, Single-Label Classification Model, Contrast Equalization, Roboflow Dataset Upload, Roboflow Dataset Upload, Polygon Zone Visualization, Stability AI Image Generation, Webhook Sink, Depth Estimation, Bounding Box Visualization, Camera Focus, Line Counter Visualization, Multi-Label Classification Model, Icon Visualization, Image Blur, Pixelate Visualization, Image Threshold, Anthropic Claude, LMM, Google Gemini, Dynamic Crop, Model Comparison Visualization, Camera Calibration, Local File Sink, Keypoint Visualization, Object Detection Model
• outputs: VLM as Detector, Byte Tracker, Google Vision OCR, SAM 3, Overlap Filter, Detections Stabilizer, Image Preprocessing, LMM For Classification, Ellipse Visualization, Triangle Visualization, Stitch Images, Detections Combine, QR Code Generator, Image Slicer, VLM as Classifier, Background Color Visualization, Model Monitoring Inference Aggregator, Segment Anything 2 Model, Template Matching, Distance Measurement, Dot Visualization, EasyOCR, Halo Visualization, Data Aggregator, Slack Notification, Byte Tracker, Color Visualization, JSON Parser, Llama 3.2 Vision, QR Code Detection, Expression, Line Counter, Size Measurement, Email Notification, Delta Filter, Corner Visualization, Mask Visualization, Continue If, Time in Zone, Roboflow Custom Metadata, Stability AI Outpainting, Barcode Detection, Dominant Color, Time in Zone, Crop Visualization, VLM as Detector, Grid Visualization, Perspective Correction, Clip Comparison, Single-Label Classification Model, Contrast Equalization, Polygon Zone Visualization, CLIP Embedding Model, Bounding Box Visualization, Camera Focus, Icon Visualization, Image Blur, Time in Zone, Path Deviation, Anthropic Claude, Cosine Similarity, Multi-Label Classification Model, Dynamic Crop, Bounding Rectangle, Path Deviation, Detections Consensus, Model Comparison Visualization, Cache Get, Local File Sink, Rate Limiter, Identify Changes, Classification Label Visualization, Circle Visualization, SIFT Comparison, Image Contours, Relative Static Crop, Detections Filter, VLM as Classifier, Stability AI Inpainting, Moondream2, Velocity, OCR Model, Florence-2 Model, SIFT, Morphological Transformation, Detections Transformation, Reference Path Visualization, Gaze Detection, SIFT Comparison, Buffer, Polygon Visualization, Florence-2 Model, Image Slicer, Detection Offset, Clip Comparison, Perception Encoder Embedding Model, Instance Segmentation Model, OpenAI, Image Convert Grayscale, Line Counter, PTZ Tracking (ONVIF).md), Object Detection Model, Keypoint Detection Model, Google Gemini, Label Visualization, Email Notification, Trace Visualization, Byte Tracker, Dynamic Zone, YOLO-World Model, OpenAI, CogVLM, Stitch OCR Detections, Detections Stitch, Cache Set, Blur Visualization, CSV Formatter, Single-Label Classification Model, OpenAI, Detections Classes Replacement, Twilio SMS Notification, Absolute Static Crop, Seg Preview, Roboflow Dataset Upload, Roboflow Dataset Upload, Property Definition, Stability AI Image Generation, Webhook Sink, Depth Estimation, Dimension Collapse, Line Counter Visualization, Instance Segmentation Model, Multi-Label Classification Model, Pixelate Visualization, Image Threshold, Detections Merge, Keypoint Detection Model, LMM, Google Gemini, Identify Outliers, Pixel Color Count, SmolVLM2, Qwen2.5-VL, First Non Empty Or Default, Camera Calibration, Keypoint Visualization, Object Detection Model

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds LMM in version v1 has.

Bindings

• input

  • images (image): The image to infer on..
  • prompt (string): Holds unconstrained text prompt to LMM mode.
  • lmm_type (string): Type of LMM to be used.
  • remote_api_key (Union[string, secret]): Holds API key required to call LMM model - in current state of development, we require OpenAI key when lmm_type=gpt_4v..

• output

  • parent_id (parent_id): Identifier of parent for step output.
  • root_parent_id (parent_id): Identifier of parent for step output.
  • image (image_metadata): Dictionary with image metadata required by supervision.
  • structured_output (dictionary): Dictionary.
  • raw_output (string): String value.
  • * (*): Equivalent of any element.

Example JSON definition of step LMM in version v1

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/lmm@v1",
    "images": "$inputs.image",
    "prompt": "my prompt",
    "lmm_type": "gpt_4v",
    "lmm_config": {
        "gpt_image_detail": "low",
        "gpt_model_version": "gpt-4o",
        "max_tokens": 200
    },
    "remote_api_key": "xxx-xxx",
    "json_output": {
        "count": "number of cats in the picture"
    }
}