Scenario has shipped a coordinated set of API, compute and web-app releases for creative teams, combining tighter model governance with new video, 3D and workflow capabilities. The three releases appeared in Scenario’s official changelog on 6 August 2026.

The update is broader than a routine model-catalogue refresh. Project administrators gain more direct control over which models a project can use, while artists and developers get additional generation options and practical changes to asset handling, automation and auditing.

Model access moves closer to the project

Scenario API v2.129.0 lets teams update the models associated with a project through the API. The matching web-app release adds project-level model controls to the interface, giving administrators a clearer way to constrain the tools available inside a production workspace.

That matters in studios where different projects have separate licensing, brand, safety or cost requirements. A curated project catalogue can reduce accidental use of an unapproved model without forcing every user to navigate the platform’s full model list.

There is also a breaking API change. Scenario removed the ipDetectionEnforcement field from project payloads and replaced it with a disableIpDetection setting that requires team-administrator privileges. Integrations that create or update projects should be checked before adopting the new API version.

Moderation becomes more selective

Scenario says its moderation pipeline should produce fewer false positives on highly stylised assets. Age and face alerts now pass through a realism classifier, while video inputs can be analysed using their first and last frames. This is a vendor-described improvement; teams with strict safety requirements should validate the behaviour against their own art styles and edge cases.

The API release also adds a query option for lighter workflow responses, a lastUsedAt timestamp on team API-key records and fixes for workflow node completion. In particular, Scenario says assets from transform and list nodes should now be exposed reliably after synchronous ForEach steps.

Those changes are less visible than a new generation model, but they address day-to-day production concerns: smaller payloads, unused-credential audits and automation runs that finish in a predictable state.

Video, 3D and audio arrive in one compute release

Compute v1.6.0 expands Scenario’s hosted catalogue. Grok Imagine Video 1.5 is available for image-to-video generation of clips up to 15 seconds at 480p or 720p, with multiple outputs per run. The release also adds MiniMax H3 video support, including text, first-and-last-frame and multimodal-reference inputs with native stereo audio.

For 3D work, the new Meshy smart-topology option can turn one image or as many as four photos into an asset with controllable polygon count, optional physically based rendering materials and A-pose or T-pose output. Scenario also added Hyper3D and Tripo v3 mesh-segmentation services.

Cartwheel integration now accepts custom character uploads for motion generation, rather than limiting users to a fixed character set. The compute release also lists new image and audio options, including Qwen Image 3.0, Seed Audio 1.0 and higher-quality music tiers. Availability does not mean every model has identical price, speed or usage rights, so production comparisons still need to be made model by model.

The web app catches up with larger assets

Scenario’s web app can now handle Gaussian splats containing as many as 20 million points, import SOG files and use a new performance mode. The release also improves lighting for models whose vertex normals are zero-filled.

The Genie assistant gains collection-management actions, allowing it to create, update or delete collections. Scenario says refreshed starter prompts, clickable model references and safeguards against runaway generation loops are intended to make those interactions easier to control.

Workflow authors get a Randomizer node for controlled variation. Users can also upload HEIC and HEIF images directly, while API-key records now show creation and last-used timestamps in the interface. Smaller fixes cover conditional-setting clutter and drag-and-drop reliability for text and image inputs.

For teams handling large 3D scenes, the higher splat ceiling is useful only if review machines and downstream tools can handle the resulting assets. Performance mode may help inside Scenario, but export size, browser memory and engine-specific limits still need to be tested across the rest of the production toolchain.

A release centred on production discipline

The common thread across the three releases is not simply more models. Scenario is pairing catalogue expansion with project scoping, credential visibility, workflow reliability and moderation changes. That combination is useful for teams moving from experiments into repeatable creative pipelines.

Existing API customers should address the IP-detection field change first. Creative teams can then evaluate the new video and 3D options using representative assets, while administrators confirm model permissions and usage costs. Scenario’s changelog does not provide a single price or availability matrix for the entire package, so the platform’s model pages and account controls remain the appropriate place to confirm current commercial terms.

A staged rollout would separate the governance changes from creative-model evaluation: update API clients and permission rules, validate existing workflows, then introduce new generation models to a limited project. That makes it easier to distinguish an integration regression from a model-quality difference when teams review the results.