TLDR
- FLUX 3 (public launch 23 July 2026): multimodal image + video + audio backbone; public video story ≈ 20 seconds in one pass, multi-shot, native audio; hosted try path on this site (BFL blog; FLUX 3 Video; Decrypt).
- MiniMax H3 Max (public speed/arena window 26 August 2026): a hosted, post-trained variant of open MiniMax H3 — not a MiniMax official product name, not the downloadable H3 weights. Public story: 5–15 second clips, up to 768p, native audio, built to return a take faster than you watch it (Design Arena; Artificial Analysis; gokayfem Odysseus demo).
- Decision rule: need many 5–15s drafts in one sitting, tight prompt follow, 480p/768p → try MiniMax H3 Max. Need a longer one-pass ~20s multi-shot with FLUX stills in the same family → FLUX 3 on this site.
- On this site: default video is FLUX 3. There is no MiniMax H3 Max generator here. H3 Max lives on minimaxh3max.com.
Key Takeaways
- These are not the same MiniMax. Open MiniMax H3 is MiniMax’s July 2026 omni-modal model: up to 15s, 2K path, native stereo, weights public (MiniMax H3; open-source note; Reuters). H3 Max is the later hosted speed variant of that base — Max weights are not public in the materials we have (Artificial Analysis).
- Clip-length public claims differ. FLUX 3 coverage centers on about 20 seconds with native audio (BFL; The Decoder). H3 Max sits on the H3 length band: 5–15 seconds (Artificial Analysis).
- Resolution ceilings are not the same knob. FLUX 3 Video ships HD (720p) with Full HD (1080p) via upscale, and later 2K/4K notes on the BFL timeline (FLUX 3 Video; @bfl_ai). H3 Max public cards stop at 768p; open H3 still owns the 2K regenerate path (AA; H3 open-source).
- Speed is the H3 Max headline, not a lab crown. Design Arena published I2V 6.4s and T2V 4.7s averages, ~50× vs the H3 base on their board (Design Arena). A follow-up clip in that thread times one sample at 23.7s vs 10m 55s for FLUX 3 Video — useful as a scout tape, not a controlled A/B we ran (thread reply). A public Odysseus demo claimed 15 seconds of video generated in 5 seconds (@gokayfem).
- Native audio is table stakes on both official stories. Listen every take (BFL; AA on H3 Max).
- Same-prompt evaluation on one hosted surface still beats a social montage. Run the prompts below on FLUX 3 video (cheap loops on Draft Mode). For the H3 Max side of the same brief, open minimaxh3max.com/video.
Head-to-head: public axes (not a ranking)
Read this table in 30 seconds. Cells are public positioning / documented features, not lab scores.
| Axis (your job) | FLUX 3 (Black Forest Labs) | MiniMax H3 Max (hosted H3 variant) | | --- | --- | --- | | Primary public pitch | Multimodal visual intelligence: image + video + audio in one architecture (BFL) | Faster hosted take on MiniMax H3: prompt follow + clips that return sooner than playback (Design Arena; minimaxh3max.com) | | What it is | BFL’s FLUX-family video model | Hosted post-trained MiniMax H3 — not MiniMax’s open H3 drop, not a MiniMax-named official SKU (AA) | | Public window | 23 July 2026 launch; Video generally available 4 August 2026 (BFL; FLUX 3 Video) | Arena/speed posts 26 August 2026; Odysseus speed demo 24 August 2026 (Design Arena; @gokayfem) | | Single-pass duration | Up to ~20 seconds (BFL; The Decoder) | 5–15 seconds (AA) | | Resolution narrative | HD + Full HD upscale; 2K/4K called out as coming on BFL channels (FLUX 3 Video; @bfl_ai) | Up to 768p on public H3 Max cards; 480p used for cheap drafts on hosts (AA; minimaxh3max.com) | | Speed narrative | Draft mode for cheap previews, then full quality (@bfl_ai Draft). One Arena sample vs H3 Max: 10m 55s (Design Arena reply) | I2V 6.4s / T2V 4.7s averages on Design Arena; ~50× vs H3 base; 15s Odysseus in ~5s (Design Arena; @gokayfem) | | Multi-shot / control | Strong launch claim: multi-shot from one prompt, keyframes, continuation, agentic chaining (BFL; FLUX 3 Video) | T2V + I2V on the hosted Max surface; omni reference/edit story stays with open H3, not Max (H3 post; AA) | | Native audio | On by default in the product story; multilingual dialogue (BFL) | Native audio on 5–15s clips (AA) | | Open weights | FLUX 3 Dev planned in the rollout; Video started as hosted/API (BFL) | Max weights not public in sources we have. Open H3 weights are a different download (H3 open-source) | | Independent board (early) | FLUX 3 Video #2 Text-to-Video Arena in an 11 August BFL/Arena note (1496 pts) (@bfl_ai) | H3 Max debut #1 Image-to-Video with audio, #3 Text-to-Video with audio on Artificial Analysis (Elo 1204 I2V, read 27 August 2026) (AA tweet; I2V board) | | Try path | Free on FLUX 3 video + image + Draft Mode | Not hosted here. Generate on minimaxh3max.com/video |
How to use the table: if the brief is “twelve variants before lunch, 15 seconds, still→motion,” weight H3 Max speed. If it is “three cuts, one character, ~20s, FLUX still language,” weight FLUX 3 on this site.
Criteria, not crowns · editorial cover for this comparison · generated for the FLUX 3 blog
Family / timeline (keep the lanes straight)
| Date | Event | Source | | --- | --- | --- | | 2026-07-23 | FLUX 3 public multimodal launch | BFL blog; Decrypt | | 2026-07-31 | MiniMax H3 (open omni-modal video) launches | MiniMax H3; Reuters | | 2026-08-03 | MiniMax H3 weights / system card public | H3 open-source | | 2026-08-04 | FLUX 3 Video generally available; Draft mode | FLUX 3 Video; @bfl_ai | | 2026-08-11 | FLUX 3 Video #2 on Text-to-Video Arena (BFL note) | @bfl_ai | | 2026-08-24 | Public 15s-in-5s Odysseus demo later attached to H3 Max | @gokayfem | | 2026-08-26 | H3 Max arena/speed window: Design Arena + Artificial Analysis | Design Arena; AA |
FLUX 3 extends Black Forest Labs’ stills line (FLUX.2) into video + audio (FLUX.2 → FLUX 3). On this site, start with What Is FLUX 3? and the multi-shot notes in FLUX 3 Video: Multi-Shot & Native Audio.
MiniMax H3 Max sits on top of MiniMax H3. Do not treat Max as “H3 but 2K,” and do not treat open H3 as “the fast hosted one.” If you want the fast hosted variant, the browser path is minimaxh3max.com — not this generator.
Community speed signals on X
These posts are illustrative. Different prompts, crops, and access tiers. They are not a score we measured on flux3.video.
- Design Arena: Pareto-style speed vs preference write-up, plus a side-by-side timing clip (main; timing reply — that reply labels the fast clip “M3 Max”; the parent thread is MiniMax H3 Max).
- Artificial Analysis: I2V #1 / T2V #3 debut note (tweet; I2V leaderboard).
- Odysseus 15s / ~5s generate (@gokayfem).
- FLUX 3 official motion language: @bfl_ai Video is here.
A hosted MiniMax H3 Max sample (Odysseus-style speed clip) is on the sister generator:
MiniMax H3 Max sample · hosted on minimaxh3max.com · not a controlled lab A/B vs FLUX 3
Scenario winners (if / then)
1. Twelve UGC drafts before lunch
If you need a pile of 9:16 or 16:9 5–15s takes the same afternoon, then weight the H3 Max speed story (Design Arena; Odysseus demo).
Try path: generate those drafts on MiniMax H3 Max. Use FLUX 3 Draft Mode on this site when you want a FLUX-family cheap preview instead.
2. Two- or three-cut ad, ~10–20 seconds, one character
If the brief is wide → close in one generation and you care that room tone sticks to the cut, then FLUX 3’s public multi-shot + ~20s + native audio story is the axis to test first (BFL; FLUX 3 Video).
On this site: write Shot 1 / Shot 2 / Shot 3 in one paragraph on video.
3. Locked product still → motion
If packaging is already approved as a still, then start image-to-video on both stacks. H3 Max’s early independent board lead is I2V with audio (AA I2V). FLUX 3 documents image-to-video and keyframes in the same family as stills (BFL).
Split try path: freeze the still on image, animate on video; run the same still on H3 Max video if you need more motion variants per hour.
4. Dialogue / lipsync explainer
If a face must speak, then both public stories include native audio — verify with headphones. FLUX 3 materials call out multilingual dialogue (BFL). H3 Max inherits H3’s audio-in-the-same-pass story at 5–15s (AA; H3). Expect retries.
On this site: Prompt B below.
5. Open weights, 2K, or omni references
If you need downloadable H3, 2K regenerate, or mixed image/video/audio references, then that is MiniMax H3, not H3 Max (H3 open-source). Max is the hosted speed lane.
On this site: you still evaluate the language of the brief on FLUX 3 video so a soft prompt does not get blamed on either model.
What We Know vs. What We Don’t
| We know (sourced) | We don’t know / won’t claim | | --- | --- | | FLUX 3 public launch 2026-07-23; Video GA 2026-08-04 (BFL; FLUX 3 Video) | A permanent global #1 between H3 Max and FLUX 3 across every job | | MiniMax H3 launch 2026-07-31; H3 Max public arena window 2026-08-26 (Reuters; Design Arena) | That MiniMax itself shipped a product named “H3 Max” — official X through 26 August still talks H3 | | H3 Max I2V Elo 1204 (#1 with audio) on Artificial Analysis the day we read the board (I2V leaderboard) | That Elo will hold next week, or that it predicts your SKU | | Design Arena speed averages and one 23.7s vs 10m 55s sample (thread) | A timed A/B we ran on flux3.video vs minimaxh3max.com | | This site hosts FLUX 3, not H3 Max | Cross-channel $/second shopping tables (not compared here) |
How to evaluate on this site (same-prompt protocol)
- Open FLUX 3 video (optional warm-up on Draft Mode).
- Run Prompt A, B, and C below unchanged (change only product names you truly need).
- Score each take (0–2 each): cut clarity, subject identity, motion realism, audio fit, crop survival.
- Keep any clip ≥7/10. Rewrite only the failing shot line.
- For the H3 Max half of the same test, paste the same three prompts on minimaxh3max.com/video — change only the host.
- Optional: lock a still on image, then image-to-video both hosts.
Prompt A — multi-shot social UGC
Vertical 9:16 UGC ad, about 10 seconds, daylight kitchen.
Shot 1 wide: young adult in a blue hoodie pours iced coffee into a clear glass, casual phone-tripod energy.
Shot 2 medium: glass fills, ice clinks, condensation visible, same hoodie and kitchen tiles.
Shot 3 close-up: satisfied sip, natural smile, soft window light, no logos.
Natural kitchen ambience and ice clink SFX, no music bed, no captions, no watermark.Prompt B — dialogue explainer beat
Horizontal 16:9 product explainer, about 8 seconds, clean home office.
Medium shot: friendly presenter in a charcoal sweater speaks to camera,
holds a small white wireless earbud case, clear enunciation, natural hand gestures.
Background: blurred bookshelf, soft daylight.
Native dialogue energy: short line about "all-day battery," ambient room tone only, no music, no on-screen text, no watermark.Prompt C — one-take street beat
Horizontal 16:9 cinematic one-take, about 12–15 seconds, golden-hour street.
Handheld gimbal push: a courier in a red jacket exits a cafe doorway, weaves past two cyclists,
then stops at a scooter, checks a phone, and rides off down a tree-lined avenue.
Continuous camera; same jacket and bag throughout; natural city ambience, scooter ignition, light traffic, no music, no logos, no watermark.
Prompt-first motion language · showcase keyframe on FLUX 3 · use with Prompt A–C above
Decision rule (one sentence)
Choose FLUX 3 on this site when you want a browser ~20s multi-shot + native-audio take in the FLUX family, and you will judge keepers yourself on our video generator (stills: image; cheap loops: Draft Mode).
Choose MiniMax H3 Max when the bottleneck is iteration speed on 5–15s, 480p/768p clips — open minimaxh3max.com / H3 Max video. Do not expect Max to be open H3, and do not expect this site to generate Max.
FAQ
Is MiniMax H3 Max the same as MiniMax H3?
No. H3 is MiniMax’s open omni-modal model (4–15s, 2K path, public weights) (H3; open-source). H3 Max is a later hosted, post-trained variant aimed at speed and prompt follow, publicly discussed from 26 August 2026 (AA). MiniMax’s own channels still describe H3, not Max, in the posts we checked that day.
Is H3 Max “better” than FLUX 3?
Not as a universal rank. Public materials show different jobs: fast 5–15s hosted drafts vs longer multi-shot native-audio generation. Better = fits the brief after the same three prompts.
How long can each generate in one pass?
FLUX 3 public coverage: about 20 seconds (BFL; The Decoder). H3 Max public cards: 5–15 seconds (AA). Host knobs can still cap what you see.
Did anyone time H3 Max against FLUX 3?
Design Arena posted one sample at 23.7s vs 10m 55s for FLUX 3 Video (reply). That is one clip in their arena, not a test we ran on these two websites. Treat it as a speed signal.
Where do I actually generate MiniMax H3 Max?
On minimaxh3max.com — start at the video generator. This site does not host that model.
Can I try FLUX 3 without installing local tools?
Yes. Open FLUX 3 video or image in the browser. Optional Draft Mode for fast previews.
Does flux3.video ship MiniMax models?
No. Default video here is FLUX 3. Sister generator for Max: minimaxh3max.com/video.
Do both models generate audio?
Yes in their public stories (BFL; AA). Native audio ≠ perfect lipsync. Headphones, every take.
Should I compare prices across platforms here?
No. This article skips $/second shopping tables. Decide on job fit and keeper rate, then price on the surface you actually ship from.
Where should I learn FLUX 3 multi-shot prompting?
Read FLUX 3 Video: Multi-Shot, ~20s Clips & Native Audio and paste Prompt A into the on-site generator.
About Avery Lin
Avery Lin is an AI model release analyst who tracks FLUX-family launches and turns public model facts into clear try paths for creators on FLUX 3.
