FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space
Source: https://arxiv.org/abs/2506.15742
KontextBench availability/licensing not checked.
Structured fields
{
"checked_at": "2026-10-17T00:00:00.000Z",
"usefulness": "Technical report behind FLUX.1 Kontext. Introduces KontextBench (1,026 image-prompt pairs across five task categories) and claims improved object/character preservation across repeated edits. Primary source for how the vendor frames and measures editing consistency.",
"source_type": "paper",
"availability": "available",
"source_owner": "Black Forest Labs",
"next_check_at": "2026-10-24T00:00:00.000Z",
"known_limitations": [
"Only the abstract page was inspected; benchmark methodology and numbers in the PDF were not read."
]
}Evidence claims
- [sourced_fact] arXiv listing: submitted 17 June 2025 (v1), revised 24 June 2025 (v2); authored by Black Forest Labs and 20 other authors. source
- [upstream_claim] Abstract claims a unified generation+editing flow-matching model with better preservation of objects and characters across iterative edits, competitive quality and faster generation, and introduces KontextBench (1,026 pairs, five task categories). source
- [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| documentation | unknown | unknown | unknown | unknown | unknown | arXiv page links to the non-exclusive distribution license 1.0; the page itself did not spell out its terms, so reuse permissions are left unknown. Only the abstract page was read, not the PDF. |