loopling.ai

Developer API

The studio's models,
from your code.

Seedance video, Astra and Claude behind one key. Paid in Ink from the wallet you already have.

Get your API key · Read the docs

Keys are issued in Settings → API. Creating one needs a paid plan or a completed Ink top-up.

What you can call

Three products, one wallet.

Every request reserves Ink up front and settles to what the vendor reports. Failures refund. The Ink is the same Ink the app spends.

Seedance video

Async generation: submit, poll, download. Seedance 2.5 up to 30 s at 1080p; 2.0 and 2.0 Fast up to 15 s.

POST /v1/video/generations

Ink per second of video

Model	Resolution	Ink / s	5 s clip
Seedance 2.5	480p	10.2	52 ($0.52)
Seedance 2.5	720p	22.9	115 ($1.15)
Seedance 2.5	1080p	56.8	284 ($2.84)
Seedance 2.0	480p	6.9	35 ($0.35)
Seedance 2.0	720p	15.1	76 ($0.76)
Seedance 2.0	1080p	37.6	189 ($1.89)
Seedance 2.0 Fast	480p	5.6	29 ($0.29)
Seedance 2.0 Fast	720p	12.1	61 ($0.61)

Aspect ratios 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 and adaptive; Seedance 2.5 tops out at 720p on 9:16 and adaptive frames.
Duration is an explicit integer: 4–30 s for 2.5, 4–15 s for 2.0 and 2.0 Fast.
References by public https URL — up to 50 for 2.5 (30 images, 10 videos, 10 audio), 12 for 2.0 (9 / 3 / 3).
Results stay for 7 days; the signed download link refreshes on every read.

Astra

GPT 6 Astra through the OpenAI Responses protocol. Point the OpenAI SDK at Loopling and keep your code.

POST /v1/responses

Ink per million tokens

Model	Input	Output	Cache read	Cache write
Astra (`gpt-6-astra`)	375	1,875	38	469

Text only in v1; streaming with SSE; function tools and structured output supported.
Long-context rates apply above 272k input tokens.

Opens after vendor certification. GET /v1/models reports which lanes are live today.

Claude

Claude Fable 5.1, Opus 5 and Sonnet 5 through the Anthropic Messages protocol. The Anthropic SDK works with a base URL.

POST /v1/messages

Ink per million tokens

Model	Input	Output	Cache read	Cache write
Claude Fable 5.1 (`claude-fable-5-1`)	1,063	5,313	27	2,125
Claude Opus 5 (`claude-opus-5`)	532	2,657	54	1,063
Claude Sonnet 5 (`claude-sonnet-5`)	213	1,063	22	425

Text only in v1; streaming, client tools, thinking and prompt caching supported.
max_tokens is required, up to 64,000; Fable runs adaptive thinking. The Anthropic SDK connects with auth_token.

Opens after vendor certification. GET /v1/models reports which lanes are live today.

One Ink, one wallet.

$1 = 100 Ink

Prices are published in Ink. Pay-as-you-go Ink costs $1 per 100, and a plan's monthly grant spends the same way. Your API keys draw on the same wallet as the app — grant first, then pay-as-you-go — and each key can carry its own cap.

Generation lanes hold the Ink for what you asked for and refund the difference when the vendor reports less.

Open Inkwell

Quickstart

Bearer auth, JSON in, JSON out. The first call below submits a clip; the SDK tabs show Astra and Claude with the client you already use.

curl https://loopling.ai/v1/video/generations \
  -H "Authorization: Bearer $LOOPLING_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.5",
    "prompt": "A paper lantern drifting over a night harbour, slow dolly in",
    "aspect_ratio": "16:9",
    "resolution": "1080p",
    "duration": 5
  }'

Questions

How do I get a key?

Sign in, open Settings → API and create one. Creating a key needs a paid plan or at least one completed Ink top-up; the key is shown once.

What does a call cost?

The Ink figure on this page: the Ink rate times the usage the vendor reports, rounded up once per call. Video reserves for the requested duration and settles to the delivered duration; text reserves for the prompt plus the output cap and settles to actual tokens.

Is this the same Ink as the app?

Yes. One wallet: the monthly grant first, then pay-as-you-go. Usage from every key shows in Settings → API next to the app's own ledger.

Can I cap what a key spends?

Each key can carry an Ink cap. When it is reached, that key answers 402 while your other keys keep working. Revoking a key refuses new calls immediately; running generations still finish and settle.

Which SDKs work?

Astra speaks the OpenAI Responses protocol, so the OpenAI SDK works with base_url set to Loopling. Claude speaks the Anthropic Messages protocol, so the Anthropic SDK works the same way. Video is plain REST with a polling loop.

What about rate limits?

60 requests per minute per key, at most 2 video generations running per account, and an optional per-key Ink cap. 429 and 503 carry Retry-After.

loopling.ai · grows with you, grows itself

hello@loopling.ai · community · pricing · Enterprise · API · privacy · terms

The studio's models,from your code.