FLUX 3 Max: Speed, Benchmarks, Pricing and API Guide

FLUX 3 Max launched on October 19, 2026. It is a post-trained version of the FLUX 3 video model, built by fal Research in collaboration with Black Forest Labs. The goal is simple: keep FLUX 3's quality and native audio while making generation fast enough to iterate like you would with images.

FLUX 3 Max is now available on PoYo through three model IDs for text-to-video, image-to-video and keyframes-to-video. The FLUX 3 Max API page includes the live generator, supported controls and pricing.

This guide covers what FLUX 3 Max changes, which speed and quality claims hold up, how it compares with FLUX 3 and Seedance 2.5, and how to call it through the PoYo API.

FLUX 3 Max at a Glance

What Is FLUX 3 Max?

FLUX 3 is the multimodal video model Black Forest Labs released in 2026. It generates up to 20 seconds of video with synchronized audio from text, images or keyframes. You can read more about it in our FLUX 3 overview.

FLUX 3 Max starts from that foundation. fal describes three layers of work behind it:

• Post-training with Black Forest Labs to improve visual fidelity, motion coherence, prompt adherence and image-to-video quality
• Inference optimization that changes how the model executes to cut latency and raise throughput
• Systems engineering that tunes the serving path for production workloads

fal reports a 50-point image-to-video Elo gain over the original FLUX 3 and calls this release the first iteration, with more post-training planned.

How Fast Is FLUX 3 Max?

The headline claim is that FLUX 3 Max is 22 times faster than Seedance 2.5 at the same quality rating. The underlying numbers are less tidy:

All of these are fal's own benchmarks for a 10-second, 16:9 clip at 720p or 768p. A later chart revised the competitor timings downward without explanation, which narrows the speed gap considerably.

The safe takeaway: FLUX 3 Max returns a 10-second clip in roughly 12 seconds under fal's test conditions. That is still very fast for a frontier video model. Real completion time depends on settings, queue load and network, so treat it as a benchmark, not a guarantee.

FLUX 3 Max Benchmarks

On the Artificial Analysis Video Arena, which ranks models by Elo from blind pairwise human votes, launch-week results were:

• Image-to-video: #2 at 1,138 Elo, statistically tied with MiniMax H3 (1,140) and Seedance 2.5 (1,138)
• Text-to-video: #4 at 1,159 Elo, tied with Seedance 2.5
• Versus base FLUX 3: ahead on both boards (FLUX 3 sat at 1,086 for image-to-video)

Artificial Analysis also reported that FLUX 3 Max scored highest of any model in the Physics capability and in the Marketing & Advertising and Live-Action Film use cases on text-to-video.

So "tied for #1" is accurate only for image-to-video, and only within confidence intervals. The clearer result is that post-training worked: Max beats its own base model on both boards.

FLUX 3 Max vs FLUX 3

Choose FLUX 3 Max for fast text-to-video, image-to-video and keyframed shots, especially when you plan to iterate on many takes. Choose FLUX 3 when you need to extend an existing clip. Both cost the same per second on PoYo for their standard workflows.

FLUX 3 Max vs Seedance 2.5

The two models are statistically tied on image-to-video quality. The practical differences are speed and workflow.

• Speed: FLUX 3 Max is several times faster on fal's benchmarks, with estimates from 5.5× to 22× depending on the chart.
• Clip length: FLUX 3 Max caps at 20 seconds. Seedance 2.5 supports longer single shots.
• References: Seedance 2.5 accepts many more reference inputs. FLUX 3 Max focuses on a start image, an optional end image or up to 10 timed keyframes.
• Control: FLUX 3 Max keyframes let you pin images to exact frame positions in the clip.

If iteration speed and keyframe timing matter most, FLUX 3 Max is the stronger fit. For long, heavily referenced shots, Seedance remains the better option.

FLUX 3 Max Pricing on PoYo

PoYo bills per second of output. Audio is included.

All three workflows use the same rates. The total cost is duration multiplied by the per-second rate.

How to Use the FLUX 3 Max API on PoYo

PoYo exposes three model IDs:

• flux-3-max/text-to-video: prompt only
• flux-3-max/image-to-video: 1–2 images in image_urls. The first sets the opening frame; an optional second guides the ending frame.
• flux-3-max/keyframes-to-video: 1–10 entries in keyframes, each with image_url and frame_index

Submit a task to POST /api/generate/submit, then poll GET /api/generate/status/{task_id} or set a callback_url.

Text-to-video example

{
  "model": "flux-3-max/text-to-video",
  "input": {
    "prompt": "One continuous shot. A barista pours steamed milk into a flat white on a wooden counter as the cup slides toward the camera. Slow push-in at counter height, shallow depth of field, warm morning window light from the left. Sound: the hiss of the steam wand fading out, a soft ceramic clink, quiet cafe chatter.",
    "duration": 10,
    "resolution": "720p",
    "aspect_ratio": "16:9",
    "sound": true
  }
}

Keyframes-to-video example

Keyframes use a 24 fps timeline, so frame_index must be unique and no greater than duration × 24. For a 10-second clip, valid indices run from 0 to 240.

{
  "model": "flux-3-max/keyframes-to-video",
  "input": {
    "prompt": "A sneaker rotates on a pedestal as studio lights shift from cool blue to warm orange. Smooth orbit camera. Sound: low ambient hum and a soft whoosh on each light change.",
    "duration": 10,
    "resolution": "1080p",
    "aspect_ratio": "1:1",
    "keyframes": [
      { "image_url": "https://example.com/front.png", "frame_index": 0 },
      { "image_url": "https://example.com/side.png", "frame_index": 120 },
      { "image_url": "https://example.com/back.png", "frame_index": 240 }
    ]
  }
}

Supported parameters

• duration: integer from 5 to 20 seconds. The default is 5.
• resolution: 720p or 1080p. The default is 720p.
• aspect_ratio: auto, 21:9, 2:1, 16:9, 4:3, 1:1, 3:4 or 9:16.
• sound: true or false. Native audio is on by default.

See the FLUX 3 Max API documentation for the full request schema and response format.

Prompt Tips for FLUX 3 Max

• One clear event per shot. Describe a single action instead of a sequence of unrelated events.
• Name the camera move. Push-in, orbit, tracking shot or static frame.
• Describe materials and light. How surfaces behave and where the light comes from.
• Write the sound. Quiet, low-motion scenes can come back nearly silent, so add an explicit audio cue.
• For image-to-video, describe only what changes. The image already sets the scene; the prompt should cover motion, camera and sound.

Best Use Cases

• Ads and product videos: strong physics and marketing scores on Artificial Analysis make it a good fit for product shots.
• Fast creative exploration: short render times let teams test many camera moves and looks per session.
• Storyboards and previsualization: keyframes pin key moments to exact times in the clip.
• Social video with sound: generate dialogue, ambience and effects in the same pass as the picture.

Frequently Asked Questions

What is FLUX 3 Max?

FLUX 3 Max is a post-trained, speed-optimized version of the FLUX 3 video model, built by fal Research with Black Forest Labs and launched on October 19, 2026. It generates 5–20 second videos with native audio from text, images or keyframes.

Is FLUX 3 Max really 22× faster than Seedance 2.5?

That figure comes from fal's launch benchmark. A later fal chart revised the Seedance timing, putting the gap closer to 5.5×. Either way, fal measures FLUX 3 Max at about 11–12 seconds for a 10-second 720p clip.

What is the difference between FLUX 3 and FLUX 3 Max?

FLUX 3 Max is faster and scores higher on image-to-video quality. The base FLUX 3 keeps more workflows, including video extension. See our FLUX 3 video guide for the base model.

How much does FLUX 3 Max cost?

On PoYo, FLUX 3 Max costs $0.17 per second at 720p and $0.29 per second at 1080p. A 10-second 720p clip costs $1.70.

How long can FLUX 3 Max videos be?

From 5 to 20 seconds per generation, in one-second steps.

Does FLUX 3 Max generate audio?

Yes. Native synchronized audio is on by default and can be turned off with sound: false.

Is FLUX 3 Max open source?

No. FLUX 3 Max is available only through APIs. The only open-weights FLUX 3 image model today is FLUX 3 Image Klein.

Final Takeaway

FLUX 3 Max makes a frontier-quality video model fast enough to iterate on. The "22×" headline depends on which chart you read, and "#1" is a statistical tie on image-to-video, but the core result holds: roughly 12-second renders, top-tier image-to-video quality and native audio, at the same per-second price as the base FLUX 3.

Try it on the PoYo FLUX 3 Max page, or read the API documentation to start integrating.

Sources: fal FLUX 3 Max launch coverage (Unite.AI), VentureBeat, TokenPost benchmark chart analysis, Black Forest Labs FLUX 3