Input billingNot collectedPlayground product / research
Output billingNot collectedNo inferred price
Specifications
Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.
Context / input limit
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Maximum input tokens
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Maximum output
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Maximum prompt characters
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Reasoning
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Tool calling
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Structured output
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Streaming
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Batch processing
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Fine-tuning
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Context caching
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Knowledge cutoff
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Release date
Not collected
Not established for this exact model or operation. Media controls, research training sizes and integration protocols do not establish language-model token or capability limits.
Do not add the 8B encoder to infer a total checkpoint size.
Architecture
Latent diffusion with EDM scheduling, a DiT image network conditioned at corresponding layers by frozen Llama3-8B, and an in-house 16-channel VAE.Source for Architecture (opens in a new tab)
Languages
Paper demonstrates English, Spanish, Filipino, French, Portuguese and Russian prompt understanding. It warns that non-English text rendering was not trained.Source for Languages (opens in a new tab)
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.
The record behind the model
Why it is included
Private hosted/research classification is inferred from the provider describing its own model and no matching weights in its reviewed official public repository inventory. Public Playground v1/v2/v2.5 weights and the frozen upstream Llama encoder are excluded from this claim. Exact weight-license terms remain unverified.
