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

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"
    }
}