FLUX.2 takes 8 references by API, 10 in Playground: Sume's range
Black Forest Labs says FLUX.2 pro, max and flex take up to 8 references by API (10 in its Playground), klein 4. Sume's input_references range is read per model.

Black Forest Labs says FLUX.2 [pro], [max] and [flex] accept up to 8 reference images through the API and up to 10 in its Playground, while [klein] accepts up to 4. On Sume, the number is whatever the model's input_references range descriptor says in GET /v1/images/models, and every reference must be a public HTTPS URL.
The limits are from the FLUX.2 overview, read 2026-10-02, and the Sume rules from the Image API.
What does Black Forest Labs publish?
The overview page lists the multi-reference limits in a table: klein up to 4, and max, pro and flex up to 8 through the API, 10 in the Playground. It also lists a maximum output resolution of 4 megapixels. The page distinguishes preview endpoints (flux-2-pro-preview, flux-2-klein-9b-preview, the latest) from pinned versions without the suffix.
| Variant | References by API | Where else |
|---|---|---|
| FLUX.2 [klein] | Up to 4 | Sub-second, 4 inference steps |
| FLUX.2 [pro] | Up to 8 | Playground up to 10 |
| FLUX.2 [flex] | Up to 8 | Playground up to 10 |
| FLUX.2 [max] | Up to 8 | Playground up to 10 |
What does Sume require?
References go in input_references as image_url objects with public HTTPS URLs. Localhost, private-network and non-HTTPS URLs are rejected before submission. A model whose descriptor is {"min": 0, "max": 0} is text-to-image only and rejects references. The legacy Image 1.0 route takes 1 to 10 URLs under image_urls, but the catalog route reads the per-model descriptor, so check it rather than assuming the vendor's 8.
How do you send references to FLUX.2 on Sume?
Use the catalog id black-forest-labs/flux.2-pro with input_references. Keep the count at or under what the descriptor prints; a request that sets a parameter the model does not list is rejected with 400 unsupported_parameter.
curl -X POST https://api.sume.com/v1/images \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "black-forest-labs/flux.2-pro",
"prompt": "Put the jacket from image 1 on the model from image 2",
"input_references": [
{"type":"image_url","image_url":{"url":"https://example.com/jacket.png"}},
{"type":"image_url","image_url":{"url":"https://example.com/model.png"}}
]
}'What should you watch for?
The Playground number is not an API number, so do not design around 10. Klein is not in Sume's repo catalog, which lists pro and flex for FLUX.2. If a request needs more references than the descriptor allows, pre-compose a contact sheet or drop the least useful ones. For the broader editing workflow see FLUX.2 multi-reference editing, and for what Sume lists for klein see the klein license and catalog post.
Reference hygiene
When a design truly needs more than the ceiling, merge references into a single board image before submitting. It loses some control, but it stays inside the contract.
- Use public HTTPS URLs; signed URLs that expire before the job runs will fail.
- Order the references in the order your prompt mentions them, and say image 1, image 2 explicitly.
- Keep each file reasonably sized; the vendor page did not state per-file limits, so this post does not either.
- Re-read the descriptor after catalog changes, since ranges are per model.
Sources
Related posts
More in Models
- flux-2-pro-preview vs flux-2-pro: which id does Sume send?
BFL has flux-2-pro-preview (latest) and flux-2-pro (fixed snapshot). Sume lists black-forest-labs/flux.2-pro and does not let you pick the BFL endpoint.
- gemini-omni-1.1-flash vs gemini-omni-flash-1.1: which id goes where
Google writes the model id gemini-omni-1.1-flash; Sume's catalog id is gemini-omni-flash-1.1. A mapping table, the old preview id, and how to avoid a typo.
- Gemini Omni audio reference: unsupported; Sume models that take audio
Google says Omni's API doesn't accept uploaded audio references. Sume's Omni row has no reference_audio_urls; Seedance 2.x, Wan 3.0 and MiniMax H3 honor audio.
- Gemini Omni extend: no new dialogue on an uploaded talking clip
Google says you can't extend an uploaded Omni clip where someone talks to add dialogue; extension only appends, to clips up to 10 s. Sume lists no extend mode.
Written by Sume