HappyOyster 1.0 Acting

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit

Not collected

Maximum input tokens

Not collected

Maximum output

Not collected

Input modalities

Text, ImageSource for Input modalities (opens in a new tab)

Output modalities

VideoSource for Output modalities (opens in a new tab)

Reasoning

Not collected

Tool calling

Not supportedSource for Tool calling (opens in a new tab)

Structured output

Not supportedSource for Structured output (opens in a new tab)

Streaming

SupportedSource for Streaming (opens in a new tab)

Live video is delivered through the SDK session; this is not a text-token SSE interface.

Batch processing

Not supportedSource for Batch processing (opens in a new tab)

Fine-tuning

Not supportedSource for Fine-tuning (opens in a new tab)

Knowledge cutoff

Not collected

Release date

2026-09-17Source for Release date (opens in a new tab)

Documented International/Global Model Studio availability, not the family’s earlier product-preview date.

Parameters

Not collected

Architecture

Not collected

Languages

Not collected

Context caching

Not supportedSource for Context caching (opens in a new tab)

Search grounding

Not supportedSource for Search grounding (opens in a new tab)

Console model playground

Not supportedSource for Console model playground (opens in a new tab)

Output resolutions

480p, 720pSource for Output resolutions (opens in a new tab)

Maximum prompt characters

2,000Source for Maximum prompt characters (opens in a new tab)

World theme / synopsis limit in characters, not tokens. Directing script beats have a separate limit.

Image input limits

One required first-frame image; strictly under 6 MB; JPG/JPEG/PNG/WebPSource for Image input limits (opens in a new tab)

Aspect ratios

9:16 (default) or 16:9; reference ratio 0.5–0.667 or 1.5–2.0 respectivelySource for Aspect ratios (opens in a new tab)

Input media transport

Image URL or base64, mutually exclusive for each imageSource for Input media transport (opens in a new tab)

Request mode

Create a world, obtain a Travel ticket, then start an SDK video sessionSource for Request mode (opens in a new tab)

World creation is asynchronous by default. A 120-second creation polling timeout is not a generation-duration limit.

Live session controls

Text instructions, pause and resume; no rewindSource for Live session controls (opens in a new tab)

Authentication

Server-side primary API key; temporary client API key and single-use Travel ticketSource for Authentication (opens in a new tab)

Authentication tokens are credentials, not billable language-model tokens.

Video duration

3-minute experience described in the product guideSource for Video duration (opens in a new tab)

The SDK ignores maxExperienceTimeSec for this mode. The guide describes the experience; this is not an independently verified universal API hard cap.

Instruction character limit

2,000Source for Instruction character limit (opens in a new tab)

Non-empty text instructions, measured in characters.

Usage metering

World creation + instruction calls where supported + generated experience secondsSource for Usage metering (opens in a new tab)

Regional original API prices; web credits and subscriptions have separate terms.

Data retention

Model Studio standard inference retention is up to 30 days, with legal/security exceptions; stateful files persist until deleted. HappyOyster is absent from the documented ZDR-supported model list. Do not treat a session ticket or artifact URL as a zero-retention guarantee.Source for Data retention (opens in a new tab)

Web product content-use policy

Web terms allow use of user content for model development and grant a broad content licenseSource for Web product content-use policy (opens in a new tab)

Separate from the Model Studio API deployment policy.

Web product retention

Purpose-based retention, including legal exceptions; no fixed general periodSource for Web product retention (opens in a new tab)

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

Why it is included

Official terms reserve rights in the models, license service access and prohibit extraction of underlying parameters or weights. The reviewed releases are hosted model services without a documented public-weight license. Public SDKs and OpenAPI documentation do not release their weights.

Lifecycle dates

Sources reviewed