FLUX 3 Video

Pricing — 1 offering(s)

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

Capability profile

Operator guidance

Use draft mode ($0.06-0.12/s) to iterate cheaply on HD previews, then `draft_enhance` the selected result at the matching standard rate rather than paying full price for exploratory generations. Choose this over seedance-2-0-bytedance.yaml when native lip-synced dialogue audio is the priority — BFL claims parity with Seedance 2.0 on image-to-video quality, but Seedance 2.0's audio story isn't confirmed to the same degree here. Choose ltx-2-3-fal.yaml instead when native audio isn't needed and cost is the primary constraint (materially cheaper per the registry's own competitor note). No independently published benchmark exists yet to arbitrate the image-to-video "tie" BFL claims against Seedance 2.0 — treat both as strong until a third party publishes a comparison.

Use cases

• Dialogue-driven scenes needing synchronized, lip-synced multilingual speech without a separate TTS/audio pass
• Image-to-video storyboarding from up to 10 keyframes, including pinned start/end frames for multi-shot sequences
• Iterative creative exploration via draft mode (cheap HD preview), then a full-quality render of the selected draft at the standard rate
• Extending an existing clip's momentum, framing, and scene logic forward via video continuation (up to 15s)

Limitations

• BFL's own comparative quality claims (text-to-video, image-to-video) are self-reported — no independently published third-party benchmark found as of this entry
• Explicitly an "initial version" per BFL's own launch post — enhanced controllability, image/video/audio-reference-combination generation, FLUX 3 Image (editing), and FLUX 3 Dev (open-weight) are named roadmap items, not yet available
• No native 4K — Full HD is an upscaled 720p render, not a native higher-resolution generation
• Video continuation (v2v) capped at 15 seconds, shorter than the 20-second ceiling for text/image-to-video
• No camera-control-specific claim (directed pan/dolly/orbit responsiveness) distinct from the multi-scene/camera-angle sequencing feature BFL describes

Citations

• FLUX 3 Video, Part 1: Generation (Black Forest Labs' own launch post)
• FLUX 3 (Black Forest Labs' own announcement of the broader multimodal model family)
• FLUX 3 Video — API documentation (Black Forest Labs' own docs)
• Pricing (Black Forest Labs' own pricing page)