Never pick a model again.Generative media stack for your agents

Build with Sequel
~/studio — claude

> four shots of a ceramic vase, warm side light

⏺ sequel - search_models(query: "cheap image model")

⎿ 3 models · flux-1-schnell $0.000634/run

⏺ sequel - get_model_schemas(model: "black-forest-labs/flux-1-schnell")

⎿ prompt, steps, width, height, seed

⏺ sequel - run_models(n: 4, steps: 4)

⎿ 4 images · $0.002536 · 4.1s

Four vessels in warm side light, generated by Sequel

⏺ Four images, $0.002536, attributed to this session.

>Describe the job…

sequel mcp · image, video, audio · ? for shortcuts

Ship generation today.100+ models, one connection, every run accounted for.

Includes

  • Semantic model search
  • Schemas on demand
  • Batch generation
  • Webhooks
  • Per-run attribution

One search across 100+ modelsranked by usecase, cost and performance

Describe the job, not the model. Search matches on capability and sorts on price, so the cheapest one that can actually do it comes back first.

poster art, the text has to be legible
search_modelstop 3 of 100+
Phoenix 1.0$0.00583/tile_512 · ~9s
Lucid Origin$0.006996/tile_512 · ~9s
FLUX.1 [schnell]$0.000053/tile_512 · ~4s

Type the joband the catalogue picks the model

No model id, no provider, no parameter names. A sentence goes in and a priced run comes back — the same three calls your agent makes.

A composer types a job — An amber glass vase lit from above, black backdrop — and Sequel picks FLUX.1 [schnell] for it, at $0.000634 a run.

Every frontier model, one integrationimage, video and audio, priced and metered per run

FLUX.1 [schnell]image · $0.000053/tile_512
GPT Image 2.5 Flareimage · $30.00/1m_tokens
Grok Imagine Videovideo · $0.05/second
MiniMax H3 Maxvideo · $0.05/second
MiniMax Speech 2.8 Turboaudio · $0.00006/character
Eleven v3audio · $0.0001/character
Eleven Turbo v2.5audio · $0.00005/character
FLUX.1 [schnell]image · $0.000053/tile_512
GPT Image 2.5 Flareimage · $30.00/1m_tokens
Grok Imagine Videovideo · $0.05/second
MiniMax H3 Maxvideo · $0.05/second
MiniMax Speech 2.8 Turboaudio · $0.00006/character
Eleven v3audio · $0.0001/character
Eleven Turbo v2.5audio · $0.00005/character
Lucid Originimage · $0.006996/tile_512
Veo 3.1video · $0.20/second
Hailuo 2.3video · $0.28/clip
Aura 1audio · $0.015/1k_characters
MiniMax Speech 2.8 HDaudio · $0.0001/character
Eleven Multilingual v2audio · $0.0001/character
Nova 3audio · $0.0052/minute
Lucid Originimage · $0.006996/tile_512
Veo 3.1video · $0.20/second
Hailuo 2.3video · $0.28/clip
Aura 1audio · $0.015/1k_characters
MiniMax Speech 2.8 HDaudio · $0.0001/character
Eleven Multilingual v2audio · $0.0001/character
Nova 3audio · $0.0052/minute
Phoenix 1.0image · $0.00583/tile_512
Wan 3.0video · $0.05/second
MiniMax H3video · $0.08/second
Grok TTSaudio · $0.000015/character
MiniMax Music 2.6audio · $0.15/track
Eleven Flash v2.5audio · $0.00005/character
Phoenix 1.0image · $0.00583/tile_512
Wan 3.0video · $0.05/second
MiniMax H3video · $0.08/second
Grok TTSaudio · $0.000015/character
MiniMax Music 2.6audio · $0.15/track
Eleven Flash v2.5audio · $0.00005/character

We built the hard half.queues, estimates and webhooks, so you never write them

Calling a model is one line. The queue, the estimate, the delivery and the audit trail are the rest — and they run here, not in your codebase.

It answers before a frame exists

A video run comes back the moment it is accepted, with the model it chose, a task id and what the render will cost. Your agent keeps its turn; the work keeps going without it.

run_modelsreturned in 0.4s
{
  "status": "in_progress",
  "task_id": "task_9f2c7a1b",
  "model": "minimax/hailuo-2.3",
  "cost": { "usd": "0.280000", "estimated": true },
  "eta_seconds": 80,
  "input": { "duration": 6, "resolution": "1080p", "fps": 24 },
  "created_at": "2026-02-11T14:02:11.482Z"
}

Priced before it starts

The estimate arrives with the task, not with the invoice. Show it, cap it, or refuse the run — all of it before the first second of compute.

minimax/hailuo-2.3
Estimatedunder cap
$0.2800
6s · 1080p · before run
Monthly cap

The webhook brings it back

A signed event lands on your endpoint when the asset is ready. Nothing polls, nothing sleeps, nothing holds a socket open to wait.

task.completeddelivered ~80s later
{
  "type": "task.completed",
  "data": {
    "task": {
      "id": "task_9f2c7a1b",
      "status": "completed",
      "assets": [{ "url": "https://cdn.sequel.sh/task_9f2c7a1b/clip.mp4" }]
    }
  }
}

Durable by id

A dropped connection costs you a notification, not the render. Redeploy, reconnect, come back tomorrow — the task is still there to fetch.

task_41d0e7c9
task_9f2c7a1bfetched 2d later
  1. Queued14:02:11
  2. Running14:02:12
  3. Completed14:03:31
task_7b31aa04

Attributed on arrival

Who asked, from which client, on which key. A batch of clips arrives as a list you can audit, bill back, or quietly stop.

MGvia mcp · Claude Codesk_live_…a4f2$0.28

Pay as you go.one balance, whichever model runs

  • One walletImage, video and audio draw from the same balance, whichever model the catalogue picks.
  • Nothing to forecastAdd funds when you want them. What you do not spend stays where it is.
  • Costed per runEvery run carries an estimate before it starts and a final figure when it lands.
Sequelpay as you go
•••• •••• ••••
Balance
$248.60
image · video · audio
one balance

One command,and every prompt gets a sequel.

Build with Sequel