Image models
Image models are evaluated for prompt adherence, anatomy, composition, rendering style, reference controls, identity consistency, resolution and adult creative range. A strong model for photorealism may be a poor fit for illustration or sequential character work, so BlinkX will describe capabilities by task rather than using a single overall score.
Confirmed models will appear here with the exact version and access status. Names under technical or commercial evaluation are not presented as available integrations.
Video models
Video evaluation adds temporal consistency, motion quality, camera control, duration, resolution, source-image fidelity and cost. Adult prompts also reveal differences in acceptance, silent modification and refusal behavior that are easy to miss in general-audience demonstrations.
BlinkX will preserve enough information to make comparisons meaningful: model version, test date, input type, prompt, relevant settings and whether the output was accepted, modified or refused.
One project across models
The purpose of a multi-model workspace is not to display a long logo wall. It is to let a creator carry the same brief and source material between appropriate models without losing the project history.
Access is opening progressively. Provider capacity, technical requirements, region and account status can affect which models are available at a given moment.