Flux 3
blackforestlabs/flux-3/image-to-videoflux-3-image-to-video- 🆕 Black Forest Labs' first multimodal Flux, here in image-to-video form
- 🎬 Animates a still input image into video with synchronized native audio
- 🧠 One multimodal model generates images, video and audio together
- 🔧 Sibling modes cover text-to-video, video-to-video, and keyframe transitions
- 💬 Multilingual dialogue plus sounds tied to on-screen physical events
- 📚 Clips can be agentically chained into multi-shot, minutes-long sequences
- 🎯 BFL reports better complex-prompt handling and text than earlier Flux
Black Forest Labs is a generative AI company based in Freiburg im Breisgau, Germany, founded by former members of Stability AI. The lab is best known for developing the Flux family of text-to-image models, which generate images from natural language prompts and have quickly…
Explore 6 more models by Flux (text-to-image model) →Flux 3 (image-to-video) is the animation entry point into Black Forest Labs' Flux 3 line: you supply a still image and a prompt, and the model produces a video clip with natively generated, synchronized audio. It sits alongside Flux 3 for pure text prompting and Flux 3 First Last Frame for controlled transitions between defined keyframes — all served from the same underlying model rather than separate specialised networks.
Unlike the still-image Flux 2 generation — Flux 2 Pro, Flux 2 Max and the editing-focused Flux 2 Max Edit — Flux 3 is described by BFL as a single multimodal model that mixes modalities and can generate images and video-with-audio jointly, from text alone or from image and video references.
BFL states that in preliminary evaluations conducted during midtraining, Flux 3 already showed significant improvement over earlier Flux versions in handling complex prompts and generating text. For video specifically, the family supports video-to-video reference transfer, generative video-audio continuation, keyframe control, high style diversity from camcorder-style footage to animation, and agentic chaining of clips into sequences lasting several minutes, with visual references used to keep characters consistent across scenes.
Audio is generated as part of the same pass rather than dubbed afterwards, covering multilingual dialogue and sound effects tied to physical events visible in the frame — a capability with no counterpart anywhere in the image-only Flux 2 family.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
Daily usage · Flux 3 · Organic sellers
| Seller | Health | Rep↓ | Sales | Video | Routing | $ / s | $ / video |
|---|---|---|---|---|---|---|---|
| Open Forge 0x1d90…b0aa | healthy | 91.11 | $55.25 | 5.4 min | #1 | $0.17 | — |
| Venice.ai Proxy 0x1f22…18c9 | healthy | 86.84 | $12.44 | 2.2 min | #2 | $0.1029 | — |
| ▲ Apex Ant 0x73b4…e736 | — | 81.71 | $0.42 | 5 s | #3 | $0.0841 | — |
| antseed-tidal-falcon-88cd 0x80d1…88cd | — | 12.78 | — | — | gated | $0.138 | — |
"Best price" and the seller table are live Antseed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = buyer trust score (0-100, the Antseed SDK's own formula). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on Antseed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.