# Perceptron Docs - [Quickstart](https://docs.perceptron.inc/perceptron-mk1/index.md): Perceptron Mk1 — vision-language model that sees and reasons across images and video. - [Changelog](https://docs.perceptron.inc/perceptron-mk1/changelog.md): Changelog - [Detect](https://docs.perceptron.inc/capabilities/detect.md): Return grounded object detections for a single image from text categories, annotated exemplars, or exhaustive detection. - [Image Q&A](https://docs.perceptron.inc/perceptron-mk1/capabilities/image-qa.md): Ask grounded questions about any scene. - [Image Captioning](https://docs.perceptron.inc/perceptron-mk1/capabilities/image-captioning.md): Generate concise or detailed descriptions for any scene. - [Object detection](https://docs.perceptron.inc/perceptron-mk1/capabilities/object-detection.md): Prompt Perceptron Mk1 to return grounded detections with normalized geometry. - [OCR](https://docs.perceptron.inc/perceptron-mk1/capabilities/ocr.md): Extract structured text with grounded prompts. - [In-context learning (Image)](https://docs.perceptron.inc/perceptron-mk1/capabilities/in-context-learning-image.md): Teach Perceptron Mk1 a new visual concept with one or many annotated examples. - [Focus](https://docs.perceptron.inc/perceptron-mk1/capabilities/focus.md): Let Perceptron Mk1 zoom into image regions via tool calls for fine-grained answers. - [Video Q&A](https://docs.perceptron.inc/perceptron-mk1/capabilities/video-qa.md): Ask grounded questions about a short video and get reasoned natural-language answers. - [Video Clipping](https://docs.perceptron.inc/perceptron-mk1/capabilities/video-clipping.md): Find the moment an event occurs in a video and return start/end timestamps. - [In-context learning (Video)](https://docs.perceptron.inc/perceptron-mk1/capabilities/in-context-learning-video.md): Show Perceptron Mk1 an example image or clip, then ask it to do the same task on a fresh video. - [Thinking](https://docs.perceptron.inc/perceptron-mk1/capabilities/thinking.md): Use reasoning traces to inspect how the model reached an answer. - [Structured outputs](https://docs.perceptron.inc/perceptron-mk1/capabilities/structured-outputs.md): Constrain model replies to Pydantic, JSON schemas or regex for reliable parsing. - [Get started](https://docs.perceptron.inc/perceptron-mk1/guides/python-sdk/getting-started.md): Install the Perceptron Python SDK, configure credentials, and send your first request. - [Authentication](https://docs.perceptron.inc/perceptron-mk1/guides/python-sdk/auth.md): Learn how to configure credentials and authenticate with the Perceptron SDK. - [Pointing basics](https://docs.perceptron.inc/perceptron-mk1/guides/python-sdk/pointing-basics.md): Understand Perceptron point, box, polygon, collection, and clip types and when to choose each for your application. - [Perceive basics](https://docs.perceptron.inc/perceptron-mk1/guides/python-sdk/perceive-basics.md): Learn the Perceptron Perceive decorator, core nodes, and the patterns for authoring your first multimodal prompt. - [Coordinate system](https://docs.perceptron.inc/perceptron-mk1/concepts/coordinates.md): Why normalized coordinates matter - [Python SDK FAQs](https://docs.perceptron.inc/perceptron-mk1/guides/python-sdk/faqs.md): Answers to the most common questions about Perceptron’s Python client. - [Files](https://docs.perceptron.inc/perceptron-mk1/guides/files.md): Upload media once and reference it across requests - [Multilook](https://docs.perceptron.inc/perceptron-mk1/guides/multilook.md): Ask many independent questions about one image or video in a single call and pay for the shared context once. - [MCP](https://docs.perceptron.inc/perceptron-mk1/guides/mcp.md): AI agents can interact with Perceptron's vision capabilities through our MCP server. - [Error messages](https://docs.perceptron.inc/perceptron-mk1/guides/error-messages.md): Diagnose and resolve common Perceptron API errors - [Batch processing](https://docs.perceptron.inc/perceptron-mk1/guides/batch.md): Process thousands of images with async workflows - [Scaling guide](https://docs.perceptron.inc/perceptron-mk1/guides/scaling.md): Meet throughput and latency goals with Perceptron Mk1 - [Tokenization guide](https://docs.perceptron.inc/perceptron-mk1/guides/tokenization.md) - [Prompting reference](https://docs.perceptron.inc/perceptron-mk1/best-practices/prompting-reference.md): Copy-paste prompts and code for each SDK primitive. - [Security](https://docs.perceptron.inc/perceptron-mk1/best-practices/security.md): Protect keys, data, and edge deployments - [Models](https://docs.perceptron.inc/perceptron-mk1/models/index.md): Vision-language models from Perceptron. - [Perceptron Mk1](https://docs.perceptron.inc/perceptron-mk1/models/perceptron-mk1.md): Best-in-class VLM with reasoning. - [Chat Completions](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/chat-completions.md): Generate a response for a text or multimodal conversation, using the OpenAI chat completions request and response format. - [Multilook Chat Completions](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/multilook-chat-completions.md): Run several independent prompts over one shared, prefilled context. - [Detect](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/detect.md): Return grounded object detections for a single image from text categories, annotated exemplars, or exhaustive detection. - [List Models](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/list-models.md): Returns a list of models available in the API. Use the `extended` query parameter to get additional metadata about each model's capabilities and supported features. - [Retrieve Model](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/retrieve-model.md): Returns a specific model by its ID. Use the `extended` query parameter to get additional metadata about the model's capabilities and supported features. - [Upload File](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/upload-file.md): Accepts a single file via multipart upload alongside a `purpose` field that declares how the file will be used. Subject to per-file and per-organization storage limits. - [List Files](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/list-files.md): Returns a paginated list of files belonging to the caller's organization. Cursor pagination via `after`; defaults match OpenAI's `/v1/files`. - [Retrieve File](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/retrieve-file.md): Returns metadata for a single file (size, content type, purpose, etc.). Use the `/v1/files/{file_id}/content` endpoint to download the bytes. - [Retrieve File Content](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/retrieve-file-content.md): Returns the raw bytes with the file's stored `Content-Type` header. - [Delete File](https://docs.perceptron.inc/perceptron-mk1/api-reference/endpoint/delete-file.md): Permanently deletes a file. Once deleted, the file can no longer be retrieved, downloaded, or referenced in requests. Returns 404 if no file with the given id exists. - [Quickstart](https://docs.perceptron.inc/isaac-0.2/index.md): Isaac 0.2 — open-weights image-based vision-language models. - [Changelog](https://docs.perceptron.inc/isaac-0.2/changelog.md): Changelog - [Detect](https://docs.perceptron.inc/capabilities/detect.md): Return grounded object detections for a single image from text categories, annotated exemplars, or exhaustive detection. - [Image Q&A](https://docs.perceptron.inc/isaac-0.2/capabilities/image-qa.md): Ask grounded questions about any scene. - [Image Captioning](https://docs.perceptron.inc/isaac-0.2/capabilities/image-captioning.md): Generate concise or detailed descriptions for any scene. - [Object detection](https://docs.perceptron.inc/isaac-0.2/capabilities/object-detection.md): Prompt Isaac 0.2 to return grounded detections with normalized geometry. - [OCR](https://docs.perceptron.inc/isaac-0.2/capabilities/ocr.md): Extract structured text with grounded prompts. - [In-context learning (Image)](https://docs.perceptron.inc/isaac-0.2/capabilities/in-context-learning-image.md): Teach Isaac 0.2 a new visual concept with one or many annotated examples. - [Focus](https://docs.perceptron.inc/isaac-0.2/capabilities/focus.md): Let Isaac 0.2 zoom into image regions via tool calls for fine-grained answers. - [Thinking](https://docs.perceptron.inc/isaac-0.2/capabilities/thinking.md): Use reasoning traces to inspect how the model reached an answer. - [Structured outputs](https://docs.perceptron.inc/isaac-0.2/capabilities/structured-outputs.md): Constrain model replies to Pydantic, JSON schemas or regex for reliable parsing. - [Get started](https://docs.perceptron.inc/isaac-0.2/guides/python-sdk/getting-started.md): Install the Perceptron Python SDK, configure credentials, and send your first request. - [Authentication](https://docs.perceptron.inc/isaac-0.2/guides/python-sdk/auth.md): Learn how to configure credentials and authenticate with the Perceptron SDK. - [Pointing basics](https://docs.perceptron.inc/isaac-0.2/guides/python-sdk/pointing-basics.md): Understand Perceptron point, box, polygon, and collection types and when to choose each for your application. - [Perceive basics](https://docs.perceptron.inc/isaac-0.2/guides/python-sdk/perceive-basics.md): Learn the Perceptron Perceive decorator, core nodes, and the patterns for authoring your first multimodal prompt. - [Coordinate system](https://docs.perceptron.inc/isaac-0.2/concepts/coordinates.md): Why normalized coordinates matter - [Python SDK FAQs](https://docs.perceptron.inc/isaac-0.2/guides/python-sdk/faqs.md): Answers to the most common questions about Perceptron’s Python client. - [Frame-by-frame tutorial](https://docs.perceptron.inc/isaac-0.2/guides/tutorials/frame-by-frame.md) - [Files](https://docs.perceptron.inc/isaac-0.2/guides/files.md): Upload media once and reference it across requests - [MCP](https://docs.perceptron.inc/isaac-0.2/guides/mcp.md): AI agents can interact with Perceptron's vision capabilities through our MCP server. - [Error messages](https://docs.perceptron.inc/isaac-0.2/guides/error-messages.md): Diagnose and resolve common Perceptron API errors - [Batch processing](https://docs.perceptron.inc/isaac-0.2/guides/batch.md): Process thousands of images with async workflows - [Scaling guide](https://docs.perceptron.inc/isaac-0.2/guides/scaling.md): Meet throughput and latency goals with Isaac 0.2 - [Tokenization guide](https://docs.perceptron.inc/isaac-0.2/guides/tokenization.md) - [Prompting reference](https://docs.perceptron.inc/isaac-0.2/best-practices/prompting-reference.md): Copy-paste prompts and code for each SDK primitive on Isaac 0.2. - [Security](https://docs.perceptron.inc/isaac-0.2/best-practices/security.md): Protect keys, data, and edge deployments - [Models](https://docs.perceptron.inc/isaac-0.2/models/index.md): Vision-language models from Perceptron. - [Isaac 0.2 2B (Preview)](https://docs.perceptron.inc/isaac-0.2/models/isaac-0.2-2b-preview.md): 2B VLM with reasoning. Sub-200ms time-to-first-token. - [Isaac 0.2 1B](https://docs.perceptron.inc/isaac-0.2/models/isaac-0.2-1b.md): Compact 1B VLM with reasoning, optimized for edge and low-latency deployments. - [Chat Completions](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/chat-completions.md): Generate a response for a text or multimodal conversation, using the OpenAI chat completions request and response format. - [Detect](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/detect.md): Return grounded object detections for a single image from text categories, annotated exemplars, or exhaustive detection. - [List Models](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/list-models.md): Returns a list of models available in the API. Use the `extended` query parameter to get additional metadata about each model's capabilities and supported features. - [Retrieve Model](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/retrieve-model.md): Returns a specific model by its ID. Use the `extended` query parameter to get additional metadata about the model's capabilities and supported features. - [Upload File](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/upload-file.md): Accepts a single file via multipart upload alongside a `purpose` field that declares how the file will be used. Subject to per-file and per-organization storage limits. - [List Files](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/list-files.md): Returns a paginated list of files belonging to the caller's organization. Cursor pagination via `after`; defaults match OpenAI's `/v1/files`. - [Retrieve File](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/retrieve-file.md): Returns metadata for a single file (size, content type, purpose, etc.). Use the `/v1/files/{file_id}/content` endpoint to download the bytes. - [Retrieve File Content](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/retrieve-file-content.md): Returns the raw bytes with the file's stored `Content-Type` header. - [Delete File](https://docs.perceptron.inc/isaac-0.2/api-reference/endpoint/delete-file.md): Permanently deletes a file. Once deleted, the file can no longer be retrieved, downloaded, or referenced in requests. Returns 404 if no file with the given id exists. - [Quickstart](https://docs.perceptron.inc/isaac-0.1/index.md): Isaac 0.1 — open-weights image-based vision-language model (legacy). - [Changelog](https://docs.perceptron.inc/isaac-0.1/changelog.md): Changelog - [Detect](https://docs.perceptron.inc/capabilities/detect.md): Return grounded object detections for a single image from text categories, annotated exemplars, or exhaustive detection. - [Image Q&A](https://docs.perceptron.inc/isaac-0.1/capabilities/image-qa.md): Ask grounded questions about any scene. - [Image Captioning](https://docs.perceptron.inc/isaac-0.1/capabilities/image-captioning.md): Generate concise or detailed descriptions for any scene. - [Object detection](https://docs.perceptron.inc/isaac-0.1/capabilities/object-detection.md): Prompt Isaac 0.1 to return grounded detections with normalized geometry. - [OCR](https://docs.perceptron.inc/isaac-0.1/capabilities/ocr.md): Extract structured text with grounded prompts. - [In-context learning (Image)](https://docs.perceptron.inc/isaac-0.1/capabilities/in-context-learning-image.md): Teach Isaac 0.1 a new visual concept with one or many annotated examples. - [Get started](https://docs.perceptron.inc/isaac-0.1/guides/python-sdk/getting-started.md): Install the Perceptron Python SDK, configure credentials, and send your first request. - [Authentication](https://docs.perceptron.inc/isaac-0.1/guides/python-sdk/auth.md): Learn how to configure credentials and authenticate with the Perceptron SDK. - [Pointing basics](https://docs.perceptron.inc/isaac-0.1/guides/python-sdk/pointing-basics.md): Understand Perceptron point, box, polygon, and collection types and when to choose each for your application. - [Perceive basics](https://docs.perceptron.inc/isaac-0.1/guides/python-sdk/perceive-basics.md): Learn the Perceptron Perceive decorator, core nodes, and the patterns for authoring your first multimodal prompt. - [Coordinate system](https://docs.perceptron.inc/isaac-0.1/concepts/coordinates.md): Why normalized coordinates matter - [Python SDK FAQs](https://docs.perceptron.inc/isaac-0.1/guides/python-sdk/faqs.md): Answers to the most common questions about Perceptron’s Python client. - [Frame-by-frame tutorial](https://docs.perceptron.inc/isaac-0.1/guides/tutorials/frame-by-frame.md) - [Files](https://docs.perceptron.inc/isaac-0.1/guides/files.md): Upload media once and reference it across requests - [MCP](https://docs.perceptron.inc/isaac-0.1/guides/mcp.md): AI agents can interact with Perceptron's vision capabilities through our MCP server. - [Error messages](https://docs.perceptron.inc/isaac-0.1/guides/error-messages.md): Diagnose and resolve common Perceptron API errors - [Batch processing](https://docs.perceptron.inc/isaac-0.1/guides/batch.md): Process thousands of images with async workflows - [Scaling guide](https://docs.perceptron.inc/isaac-0.1/guides/scaling.md): Meet throughput and latency goals with Isaac 0.1 - [Tokenization guide](https://docs.perceptron.inc/isaac-0.1/guides/tokenization.md) - [Prompting reference](https://docs.perceptron.inc/isaac-0.1/best-practices/prompting-reference.md): Copy-paste prompts and code for each SDK primitive on Isaac 0.1. - [Prompting tips](https://docs.perceptron.inc/isaac-0.1/best-practices/prompting-tips.md): Techniques to improve accuracy, reduce verbosity, and get better results. - [Security](https://docs.perceptron.inc/isaac-0.1/best-practices/security.md): Protect keys, data, and edge deployments - [Models](https://docs.perceptron.inc/isaac-0.1/models/index.md): Vision-language models from Perceptron. - [Isaac 0.1](https://docs.perceptron.inc/isaac-0.1/models/isaac-0.1.md): Legacy 2B VLM, still supported for existing integrations. - [Chat Completions](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/chat-completions.md): Generate a response for a text or multimodal conversation, using the OpenAI chat completions request and response format. - [Detect](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/detect.md): Return grounded object detections for a single image from text categories, annotated exemplars, or exhaustive detection. - [List Models](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/list-models.md): Returns a list of models available in the API. Use the `extended` query parameter to get additional metadata about each model's capabilities and supported features. - [Retrieve Model](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/retrieve-model.md): Returns a specific model by its ID. Use the `extended` query parameter to get additional metadata about the model's capabilities and supported features. - [Upload File](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/upload-file.md): Accepts a single file via multipart upload alongside a `purpose` field that declares how the file will be used. Subject to per-file and per-organization storage limits. - [List Files](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/list-files.md): Returns a paginated list of files belonging to the caller's organization. Cursor pagination via `after`; defaults match OpenAI's `/v1/files`. - [Retrieve File](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/retrieve-file.md): Returns metadata for a single file (size, content type, purpose, etc.). Use the `/v1/files/{file_id}/content` endpoint to download the bytes. - [Retrieve File Content](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/retrieve-file-content.md): Returns the raw bytes with the file's stored `Content-Type` header. - [Delete File](https://docs.perceptron.inc/isaac-0.1/api-reference/endpoint/delete-file.md): Permanently deletes a file. Once deleted, the file can no longer be retrieved, downloaded, or referenced in requests. Returns 404 if no file with the given id exists. ## OpenAPI Specs - [openapi](/api-reference/openapi.json)