Should I replace Convex with Pixeltable?

Not if you picked Convex for reactivity. Pixeltable has no realtime push in this benchmark. Many teams keep Convex for the app and put media and retrieval in Pixeltable.

Why is Convex’s line count so high?

429 app lines, 681 with compute-service. Most of the app tax is two things this contract imposes: an action cannot write to the database (videos.ts), and we asked for REST (http.ts). A reactive-client implementation would not pay either.

Is Pixeltable faster?

All three beat realtime on this laptop. Convex ingests in 53.7s against Pixeltable’s 66.7s — about 13 seconds on ten minutes of footage — and transcript search is 12ms vs 19ms. Most of search is embedding the query. This is CPU and local models, not Cloud. The durable Pixeltable claims are structural: where media processing runs, what happens when you add a column, and that processing fires for any writer.

Pixeltable vs Convex

Same video-intelligence app, two implementations. Pick Convex when reactivity is the point — the UI re-renders on write. Pick Pixeltable when the pipeline is the product. A REST-shaped benchmark is Convex’s worst event; discount this column accordingly.

pip install 'pixeltable[serve]'

See how it works

The trade

Side	Pixeltable	Convex
At a glance	129 lines, one file: the pipeline is the schema ffmpeg, Whisper, and CLIP run in-process; no second service Declared HTTP routes; a missing query is a 422 before any handler runs Adding a column backfills in place; processing fires for any writer	429 app lines plus 252 in compute-service, and the easiest install of the three npx convex dev: anonymous local backend, no account, no Docker 109 lines in videos.ts because an action cannot write to the database 90 lines in http.ts because this contract asked for REST instead of the reactive client

What we measured

Same app. Convex wins install, ingest, and transcript search. Pixeltable wins in-platform media. REST is Convex’s worst event.

Feature	Pixeltable	Convex
Local install	pip install; large Python deps (torch, whisper, sentence-transformers)	npx convex dev — no account, no Docker, easiest of the three
Total code for this app	129 lines, 1 file	681 lines (429 + 252 compute-service), 7 files
ffmpeg, Whisper, CLIP	Computed columns, same process	compute-service; the Convex runtime cannot run them
Writes from an action	Insert is a row; computed columns run	An action cannot write; 109 lines of mutations in videos.ts
HTTP API	add_query_route derives the signature; malformed requests are 422	Five http.route blocks. Validators sit inside the function; a failure is a 500
Reactive client	None here — you would write polling or a websocket	The reason most teams pick Convex; discarded by this REST contract
Vector search result limit	No ceiling in this implementation	vectorSearch clamps to 256
Ingest, 20 videos / 10 min	66.7s, 9.04× realtime; first video 5.3s	53.7s, 11.25× realtime; first video 3.2s
Transcript search p50	18.5ms	11.8ms
Add a derived column (lines)	1 line, 1 file	53 lines, 2 files; reverting needs a second migration
Processing for any writer	Yes — the pipeline is the schema	No, unless you add a scheduled action
Vendor checker in CI	ruff only	@convex-dev/eslint-plugin and tsc --noEmit against generated code

All three beat realtime on this laptop

20 videos, 10 minutes of footage, CPU, local models — not Cloud. Bold is best on that row.

Ingest

Feature	Pixeltable	Convex
Wall time	66.7s	53.7s
Faster than realtime	9.04x	11.25x
First video	5.3s	3.2s
Median video	3.1s	2.4s

Search

Feature	Pixeltable	Convex
Frame search p50	39.0ms	26.6ms
Frame search p95	42.9ms	40.7ms
Transcript search p50	18.5ms	11.8ms
Transcript search p95	21.6ms	17.7ms

20 videos, 603.7 seconds of footage, 603 frames, 70 transcript chunks, one laptop, CPU, local models (Pixeltable 0.7.8). All three finished 20 of 20, all faster than realtime. Supabase led ingest (11.64× vs Pixeltable’s 9.04×); frame search is 26–39ms and most of that is embedding the query, not the index. 603 vectors is not a million-row benchmark, hosted Pixeltable Cloud was not in this run, and none of this is a cost comparison.

Adding a column to live data

One computed title-embedding index on a populated catalog. Lines compound; these seconds do not.

Feature	Pixeltable	Convex
Schema change	3.88s	5.98s
Backfill	same step	1.39s
Total	3.88s	7.37s
Lines written	1	53
Files touched	1	2

Supabase is the fastest in wall time. At two dozen rows the backfill is noise; anyone quoting these seconds as a scaling result is quoting noise.
Reverting is not symmetric. Convex needs a second migration (17 of its 53 lines) because pushing a schema that no longer declares the field is rejected while documents still carry it.
After pxt schema update, an insert against the already-registered route answers 409 until pxt service update. Reads keep working. The other two resolve the table on every request.
Backfill time is the part that scales, and this corpus cannot show it. Pixeltable’s backfill is work proportional to the rows that changed; a backfill script is work proportional to the table.

Ingest a video

Insert a video. Frames, audio, transcripts, embeddings, and scenes have to exist after that. On Pixeltable they are the schema. On the other two they live in the ingest path and in a second service.

Pixeltable

class Videos(TableModel, name='videos'):
    video: pxt.Video
    title: pxt.String
    audio = extract_audio(video, format='mp3')
    duration_sec = pxtf.video.get_duration(video)
    scenes = video.scene_detect_content(threshold=8.0)

class Frames(TableModel, name='frames', base=Videos,
             iterator=frame_iterator(Videos.video, fps=1.0)):
    still = pxtf.image.resize(frame, (320, 180))
    __indexes__ = [pxt.EmbeddingIndex(frame, embedding=VISUAL)]

class Chunks(TableModel, name='chunks', base=Videos,
             iterator=audio_splitter(Videos.audio, duration=10.0)):
    transcript = transcribe(audio_segment, model='base.en').text.astype(pxt.String)
    __indexes__ = [pxt.EmbeddingIndex(transcript, embedding=SEMANTIC)]

Videos.insert([{'video': 'lecture.mp4', 'title': 'CS101'}])

Convex

const { frames } = await compute("/extract-frames", { video_url, fps: FRAME_FPS });
const { embeddings } = await compute("/embed-clip", { images_b64: frames });
const frameRows = await Promise.all(frames.map(async (b64, i) => ({
  frameIdx: i,
  imageStorageId: await ctx.storage.store(new Blob([decodeBase64(b64)])),
  embedding: embeddings[i],
})));
await ctx.runMutation(internal.videos.insertFrames, { videoId, rows: frameRows });

Search frames

Find frames of a whiteboard. Pixeltable asks the index. The other two embed the query themselves, then join or fetch rows in a second step.

Pixeltable

sim = Frames.frame.similarity(string=query)
return (
    Frames.order_by(sim, asc=False)
    .limit(limit)
    .select(
        frame_url=Frames.still,
        frame_idx=Frames.pos,
        video_title=Frames.title,
        similarity=sim,
    )
)

Convex

const { embeddings } = await compute("/embed-clip", { texts: [query] });
const hits = await ctx.vectorSearch("frames", "by_embedding", {
  vector: embeddings[0],
  limit: clamp(limit), // vectorSearch is 1-256
});
return await ctx.runQuery(internal.search.framesByIds, {
  ids: hits.map((h) => h._id),
  scores: hits.map((h) => h._score),
});

Serve over HTTP

Expose search over HTTP. Pixeltable derives the route from the query. Supabase puts five paths in one function. Convex writes five REST routes only because this contract asked for REST.

Pixeltable

api = FastAPIRouter(name='api')
api.add_insert_route(Videos, path='/videos', inputs=[Videos.video, Videos.title], background=True)
api.add_query_route(path='/videos', query=list_videos, method='get')
api.add_query_route(path='/search/frames', query=search_frames, method='post')
api.add_query_route(path='/search/transcripts', query=search_transcripts, method='post')

Convex

http.route({
  path: "/search/frames",
  method: "POST",
  handler: httpAction(async (ctx, req) =>
    guarded(async (r) => {
      const body = await readJson(r);
      return json(await ctx.runAction(api.search.searchFrames, {
        query: requireString(body.query, "query"),
        limit: readLimit(body.limit),
      }));
    })(req)
  ),
});

When to choose which platform

Choose Pixeltable when

The pipeline is the product
Media in, models and retrieval out, in one Python file. Processing belongs to the table, so a row written from a shell is processed the same way as a row written over HTTP.
You do not want to operate a second runtime for ffmpeg
The Convex runtime cannot execute ffmpeg. Three compute-service endpoints have no hosted-API substitute. That extra service is most of the orchestration hops.

Choose Convex when

Reactivity is the point
Build the same app with Convex’s reactive client instead of five REST endpoints and http.ts disappears along with both taxes. The client re-renders on write for free, and mutations are transactional.
You want the easiest local backend
npx convex dev gives a working local backend with no account and no Docker. That is the easiest install of the three, and it is not close.

Making the right choice

Discount the REST column
- videos.ts (109 lines) and http.ts (90 lines) are taxes this contract imposes, not Convex’s native shape.
- Reproduce the numbers: https://github.com/pixeltable/pixeltable-vs-supabase-vs-convex

Frequently asked questions

More comparisons

One file. The whole pipeline.

Declare the tables. Apply the schema. Insert a row. Serve the same file.

pip install 'pixeltable[serve]'

See how it works Get expert guidance

Pixeltable vs Convex

The trade

What we measured

All three beat realtime on this laptop

Ingest

Search

Adding a column to live data

Ingest a video

Search frames

Serve over HTTP

When to choose which platform

Choose Pixeltable when

Choose Convex when

Making the right choice

Discount the REST column

Frequently asked questions

Should I replace Convex with Pixeltable?

Why is Convex’s line count so high?

Is Pixeltable faster?

More comparisons

One file. The whole pipeline.