APIs
Broad Search
Input: A text queryOutput: Ranked web results for each sub-query, with relevant highlights and optional full contentBest for: Comparisons, research, topic surveys, multi-angle questionsHow to use: Feed the grouped results to a model to answer the whole question at once.
Web Search
Input: A text queryOutput: Ranked web results with relevant highlights, optional full contentBest for: Search bars, answer engines, news lookup, market or company trackingHow to use: Show the results directly, or pass the results to a model to generate grounded answers.
Image Search
Input: A text query and/or an imageOutput: Ranked image results; for design topic, also a summary and reusable snippet.Best for: Image lookup, visual references, design assets and inspirationHow to use: Pass the results to a model as visual or design reference.Access: In beta. Contact us to request access.
Video Search
Input: A text queryOutput: Ranked video results with timestamped segment matching the queryBest for: Video lookup, clip discovery, media previewsHow to use: Pass the results to a model to reference the matching video.Access: In beta. Contact us to request access.
Extract
Input: One or more URLsOutput: Clean page content with optional highlights, page classification, and media resourcesBest for: Reading pages, cleaning articles, knowledge ingestionHow to use: Store, display, summarize, or pass the content to a model as context.
Model
Input: A question, chat messages, or an image descriptionOutput: Model responses or generated imagesBest for: Chat assistants, agents, image generation, workloads that need fresh web knowledgeHow to use: Use the responses to build chat, agent, or image features.
Embedding
Input: TextOutput: Text vectorsBest for: Semantic search, RAG, recommendationsHow to use: Convert text into vectors, store them in a vector database, and compare queries against vectors.
VL Embedding
Input: Text, images, videosOutput: Multimodal vectorsBest for: Cross-modal search, visual retrievalHow to use: Convert multimodal content into vectors so text, images, and videos can be searched or matched.
Applications
Use applications when you want Octen to run a complete workflow instead of calling each API step yourself.Answer
Input: A question or chat messagesOutput: A synthesized answer grounded in search results, with the sub-queries and sources usedBest for: Q&A, chat assistants, grounded answers with citationsHow to use: Send the user’s question. Octen returns the synthesized answer and its sources.
Deep Research
Input: A research questionOutput: A structured report with supporting evidenceBest for: Deep dives, strategic research, technical investigations, trend analysisHow to use: Enter a research question. Octen plans, searches over multiple rounds, and streams the final report.
Multimodal Chat
Input: A question or chat messagesOutput: An interleaved response of text, images, and videos, grounded in multimodal search resultsBest for: Visual Q&A, rich chat experiencesHow to use: Send the user’s question. Octen replies with interleaved text, images, and videos.Access: In beta. Contact us to request access.
Grounded Generation
Input: A generation promptOutput: Generated images or videos grounded in search resultsBest for: Fact-grounded visuals, illustrated explainersHow to use: Send a prompt. Octen generates grounded images or videos.Access: In beta. Contact us to request access.
Common Patterns
- Search and answer: use Broad Search to cover a query from multiple angles, then pass the results into your LLM or agent.
- Design reference: use Image Search with the design topic, then pass the structured summary and HTML snippet to your model as design context.
- Fetch and read: use Extract when you already know the URLs and need the page content.
- Chat and generate: use the Model API to chat or generate images directly, and turn on built-in search tools when answers need live web data.
- Research and report: use Deep Research when you need a structured report with supporting sources.
- Discover and create: use Grounded Generation to create images and videos based on real web content.