Supported Data Types

Instill Core Pipelines support a wide range of structured and unstructured data types for ingestion, transformation, and processing. This includes both primitive JSON-compatible values and media-rich files such as documents, images, video, and audio.

For a conceptual overview, see Instill Types.

Primitive Data Types

These are basic types supported in structured pipelines and API interactions:

• string – UTF-8 encoded text
• number – Integer or floating-point number
• boolean – true or false
• json – JSON object or array

These types are typically used in parameter inputs, model configuration, or pipeline inputs and outputs.

Document

Instill Core supports a wide range of text-based formats used for documents, web pages, and structured tables.

• .doc – Microsoft Word 97–2003
• .docx – Microsoft Word 2007–2019
• .ppt – Microsoft PowerPoint 97–2003
• .pptx – Microsoft PowerPoint 2007–2019
• .xls – Microsoft Excel 97–2003
• .xlsx – Microsoft Excel 2007–2019
• .html – HTML document
• .txt – Plain text
• .md – Markdown
• .csv – Comma-separated values
• .pdf – Portable Document Format

Image

Image formats used in OCR, classification, and other Vision AI tasks.

• .jpeg – JPEG image
• .png – PNG image
• .gif – GIF image
• .webp – WebP image
• .tiff – TIFF image
• .bmp – Bitmap image

Audio

Audio file types supported for speech recognition and sound analysis.

• .mp3 – MP3 audio
• .wav – WAV audio
• .aac – AAC audio
• .ogg – Ogg Vorbis
• .flac – FLAC audio
• .m4a – MPEG-4 audio
• .wma – Windows Media Audio
• .aiff – AIFF audio

Video

Video formats for transcription, frame analysis, and multimedia workflows.

• .mp4 – MPEG-4 video
• .avi – AVI video
• .mov – QuickTime video
• .webm – WebM video
• .mkv – Matroska video
• .flv – Flash Video
• .wmv – Windows Media Video
• .mpeg – MPEG video