Every model that bild offers from Sourceful (provider ID sourceful, API key variable SOURCEFUL_API_KEY), with every option each model accepts and the constraints the model declares. An option the provider requires opens with Required.; every other option is optional. bild info sourceful/<model> prints the same facts for the binary you run. What the constraints mean is described in Parameter adjustment, the flags in Generation flags, and how to name a model in Model specifiers.
4 image models, 0 video models.
Contents
Riverflow 2.5 Pro
| Name | Medium | Aliases | Full ID |
|---|---|---|---|
| Riverflow 2.5 Pro | image | riverflow, riverflow-v2.5-pro, riverflow-pro |
sourceful/riverflow-2.5-pro |
| Option | Constraints |
|---|---|
--aspect-ratio |
Allowed values: auto, 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16 |
--resolution |
Allowed values: 1K, 2K, 4K |
--output-format |
Allowed values: webp, png, jpg, jpeg |
--thinking-level |
Allowed values: low, medium, high, xhigh |
--input-media |
Repeat maximum: 10 |
--background |
Allowed values: original, transparent |
--prompt-upsampling |
Riverflow 2.5 Fast
| Name | Medium | Aliases | Full ID |
|---|---|---|---|
| Riverflow 2.5 Fast | image | riverflow-v2.5-fast |
sourceful/riverflow-2.5-fast |
| Option | Constraints |
|---|---|
--aspect-ratio |
Allowed values: auto, 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16 |
--resolution |
Allowed values: 1K, 2K, 4K |
--output-format |
Allowed values: webp, png, jpg, jpeg |
--thinking-level |
Allowed values: low, medium, high, xhigh |
--input-media |
Repeat maximum: 10 |
--background |
Allowed values: original, transparent |
--prompt-upsampling |
Riverflow 2 Fast
| Name | Medium | Aliases | Full ID |
|---|---|---|---|
| Riverflow 2 Fast | image | none | sourceful/riverflow-2-fast |
Generate images from text or edit reference images using Riverflow 2 Fast.
| Option | Constraints |
|---|---|
--aspect-ratio |
Allowed values: auto, 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16; Use auto to infer the ratio from the prompt or reference image. The direct API documents 1:1 as its default. |
--resolution |
Allowed values: 1K, 2K; Fast supports 1K and 2K only. The shared API schema also lists 4K, but the direct Fast API rejects it; 4K is supported by Pro. |
--input-media |
Repeat maximum: 4; Accepts local images or public HTTPS image URLs, preserving their order. Local images are sent as data URIs; URL inputs remain URLs. Excess references are omitted with a notice. Frame prefixes are removed with a notice. Fast accepts four references, although the shared API schema allows ten. The direct API enforces four, matching Sourceful's model-specific documentation. Sourceful documents a 4.5 MB request-body limit, but Fast accepted approximately 5.5 MB and 22 MB requests on 2026-10-05. The actual maximum is unverified. Bildomat forwards the request and reports any provider refusal; public URLs avoid embedding large image data. |
--prompt-upsampling |
When enabled, the API enhances the prompt using use-case context. The API default is false. Prompts must contain at least two characters; the direct API documentation says text beyond 24000 characters is truncated before processing. |
Riverflow 2 Pro
| Name | Medium | Aliases | Full ID |
|---|---|---|---|
| Riverflow 2 Pro | image | none | sourceful/riverflow-2-pro |
Generate images from text or edit reference images using Riverflow 2 Pro.
| Option | Constraints |
|---|---|
--aspect-ratio |
Allowed values: auto, 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16; Use auto to infer the ratio from the prompt or reference image. The direct API documents 1:1 as its default. |
--resolution |
Allowed values: 1K, 2K, 4K; Pro supports 1K, 2K, and 4K. |
--input-media |
Repeat maximum: 10; Accepts local images or public HTTPS image URLs, preserving their order. Local images are sent as data URIs; URL inputs remain URLs. Excess references are omitted with a notice. Frame prefixes are removed with a notice. Pro accepts ten references. Sourceful documents a 4.5 MB request-body limit and recommends URLs for large images. Bildomat forwards the request and reports any provider refusal; Pro's actual maximum request size has not been verified. |
--prompt-upsampling |
When enabled, the API enhances the prompt using use-case context. The API default is false. Prompts must contain at least two characters; the direct API documentation says text beyond 24000 characters is truncated before processing. |
Bildomat 0.0.7Revised 2026-10-06