--- title: "Pixeltable and Mixpeek" description: "Mixpeek is a managed multimodal retrieval API (your bucket, from $25/mo plus usage). Pixeltable is one app file you own. Last reviewed Oct 2026." keywords: - Mixpeek alternative - Pixeltable vs Mixpeek - video search API - multimodal retrieval - managed video search url: "https://www.pixeltable.com/compare/pixeltable-vs-mixpeek" --- # Pixeltable and Mixpeek Updated: 2026-10-23 ## Which is better, Pixeltable or Mixpeek? It depends. Mixpeek belongs when you want a managed multimodal retrieval API: connect a bucket, extract scenes, speech and OCR, and search over HTTP without running models. Pixeltable belongs when the clip, the timestamp, the index and HTTP should live in one app.py you own. Keep Mixpeek if the hosted retriever is the product. Pixeltable Cloud is Limited Beta. As of 2026-10-23. Mixpeek is a managed retrieval API: connect a bucket, extract scenes, speech and OCR, and search over HTTP. Pixeltable is the application schema those results can live on. Often complementary, not exclusive: call Mixpeek when the hosted retriever is the product, and keep the clip in a Video column you own when the table is the product. ## Summary ### Pixeltable - Video column plus a frame_iterator view; insert indexes new clips - CLIP EmbeddingIndex on frames; optional Whisper on the soundtrack - column.similarity(string=) is the score; the query returns a title, timestamp, and score - Same file can serve FastAPIRouter; Cloud is Limited Beta ### Mixpeek - Managed indexing: point at video, audio, or documents in your bucket - Retrievers.execute over HTTP; Python SDK pip install mixpeek - BYO vectors (MVS) or managed extractors (features such as video_search) - BYO-Cloud self-host option; Build plan from $25/mo plus usage (mixpeek.com, 2026-10-23) ## Comparison | Feature | Pixeltable | Mixpeek | | --- | --- | --- | | What it is | Application schema: store, transform, index, serve | Managed multimodal retrieval API | | Where the clip lives | A Video column you query, version, and attach other columns to | An existing bucket Mixpeek can read in place, or a Mixpeek bucket you upload into | | Who runs the models | You, locally or on Cloud (Limited Beta) | Mixpeek, unless you bring your own vectors or deploy BYO-Cloud | | Insert a new file | Insert a row; frames and the index run for that clip | Upload or connect the object, trigger the collection, poll the task | | Search API | column.similarity(...) on the frame view you own | retrievers.execute against a hosted retriever | | Images, PDFs, HTTP you own | Same schema: Image, Document, FastAPIRouter | Same API family for video, image, audio, documents; not an application table | | Self-host | pip install pixeltable; Cloud is Limited Beta | Your own cloud (BYO-Cloud) (mixpeek.com FAQ, accessed 2026-10-23) | ## Index a folder of videos and search by text Pixeltable: frame_iterator view and CLIP EmbeddingIndex. Apply with pxt schema update app.py video_lib. Mixpeek: the quickstart creates a Mixpeek bucket, uploads into it, triggers the collection once, polls tasks.get, creates a retriever, then retrievers.execute with inputs query_text. Connecting an existing bucket is a separate FAQ path. Snippets from docs.mixpeek.com/docs/overview/quickstart (accessed 2026-10-23); not executed here (needs a Mixpeek API key and plan). ### Pixeltable ```python import pixeltable as pxt from pixeltable.functions.huggingface import clip from pixeltable.functions.video import frame_iterator from pixeltable.serving import FastAPIRouter TableModel = pxt.model_base() clip_embed = clip.using(model_id='openai/clip-vit-base-patch32') class Videos(TableModel, name='videos'): title: pxt.String video: pxt.Video class Frames( TableModel, name='frames', base=Videos, iterator=frame_iterator(video=Videos.video, fps=1), ): __indexes__ = [pxt.EmbeddingIndex(frame, embedding=clip_embed)] @pxt.query def search_video(query: str, limit: int = 5): sim = Frames.frame.similarity(string=query) return ( Frames.order_by(sim, asc=False) .limit(limit) .select(title=Videos.title, frame_attrs=Frames.frame_attrs, score=sim) ) api = FastAPIRouter(name='video_lib') api.add_insert_route( Videos, path='/videos', inputs=[Videos.title], uploadfile_inputs=[Videos.video] ) api.add_query_route(path='/search', query=search_video) ``` ### Mixpeek ```python import os from mixpeek import Mixpeek client = Mixpeek(api_key=os.environ['MIXPEEK_API_KEY']) ns = client.namespaces.create(namespace_name='video-demo') client.namespace = ns['namespace_id'] bucket = client.buckets.create( bucket_name='demo-videos', bucket_schema={'properties': {'video': {'type': 'video'}}}, ) # This quickstart creates a Mixpeek bucket and uploads into it. # An existing S3, GCS, or Azure bucket is the separate FAQ path. col = client.collections.create( collection_name='video-scenes', source={'type': 'bucket', 'bucket_ids': [bucket['bucket_id']]}, features=['video_search'], ) client.buckets.upload( bucket['bucket_id'], blobs=[{ 'property': 'video', 'type': 'video', 'data': 'https://example.com/lecture.mp4', }], ) result = client.collections.trigger(col['collection_id']) task = client.tasks.get(task_id=result['task_id']) # Repeat tasks.get until task['status'] is COMPLETED or COMPLETED_WITH_ERRORS. # Do not call collections.trigger again to poll. ret = client.retrievers.create( retriever_name='video-search', input_schema={'query_text': {'type': 'text', 'required': True}}, collection_identifiers=[col['collection_id']], stages=[{ 'stage_name': 'search', 'stage_type': 'filter', 'config': { 'stage_id': 'feature_search', 'parameters': { 'searches': [{ 'feature_uri': 'mixpeek://universal_extractor@v1/gemini-embedding-2', 'query': {'input_mode': 'text', 'value': '{{INPUT.query_text}}'}, 'top_k': 50, }], 'final_top_k': 20, }, }, }], ) results = client.retrievers.execute( ret['retriever_id'], inputs={'query_text': 'person at a whiteboard'}, ) for doc in results['documents']: print(doc['document_id'], doc.get('start_time'), doc['score']) ``` ## When to choose Pixeltable - **You need the clip next to other columns**: Transcripts, detections, documents, and HTTP in one file. Mixpeek can still be the retriever you call from a computed column or a route. - **A coding agent writes the app**: One app.py to review. pip install pixeltable. Pixeltable Cloud is Limited Beta, not GA. ## When to choose Mixpeek - **A managed video search API is the product**: You want extraction and retrieval over HTTP without running CLIP or Whisper. Pixeltable is extra if the API response is enough. - **You already pay for Mixpeek**: Keep it. Point Pixeltable at the same bucket when you also need an owned table; otherwise stay on their SDK. Docs: https://www.mixpeek.com/docs/overview/quickstart ## Making the right choice ### And, not or - Do not migrate off Mixpeek just to complete the stack. Keep it if the hosted retriever is the contract. - The homepage FAQ says Mixpeek can read an existing S3, GCS, Azure, or S3-compatible bucket in place (accessed 2026-10-23). The code sample uses the quickstart: a Mixpeek bucket you upload into. - Managed path: namespaces.create, buckets.create, collections.create with features, upload, trigger, retrievers.execute (quickstart, accessed 2026-10-23). - Build starts at $25/mo plus usage for both MVS and Managed (mixpeek.com pricing, accessed 2026-10-23). Confirm current rates on their site. ## FAQ ### Is Pixeltable a Mixpeek alternative? Not as a managed extraction and retrieval API. Mixpeek is that API. Pixeltable is the table those results can live on, plus CLIP, Whisper, and HTTP you own. If you only need hosted search over a bucket, Mixpeek is the product. ### Does Pixeltable replace Mixpeek extractors? No. Mixpeek ships managed extractors (scenes, speech, OCR, faces) as the product. Pixeltable calls models you install (CLIP, Whisper) as columns. Different ownership, not a drop-in extractor swap. ### Can I self-host either side? Pixeltable: pip install locally; Cloud is Limited Beta. Mixpeek: your own cloud (BYO-Cloud), plus a self-hosted option on the homepage FAQ (accessed 2026-10-23). Mixpeek's FAQ lists no SOC 2 or HIPAA certification (accessed 2026-10-23). ### Where were these facts checked? Last reviewed: Oct 2026. Mixpeek: mixpeek.com (product, pricing from $25/mo plus usage, BYO-Cloud), mixpeek.com/video-search (Video Search API, retrievers.execute), docs.mixpeek.com/docs/overview/quickstart (pip install mixpeek, namespaces, buckets, features=["video_search"]). Pixeltable: docs.pixeltable.com/howto/cookbooks/video/video-extract-frames and /howto/deployment/serving.