中

Which LoRA Acceleration Is Right for You? The Blind-Test Leaderboard

Updated 2026-09-20 · Leaderboard source: H3 Acceleration Arena (by multimodalart, Hugging Face Space) · fetched 2026-09-28

TL;DR: as of 2026-09-28, the i2v slice is topped by LightX2V 8-step (native profile) and the t2v slice by larry v4 — the only entry in the top three of both slices; the plain 28-step baseline still ranks #2 on t2v. “Fastest” and “best looking” are not the same leaderboard.

Where this leaderboard comes from

H3 Acceleration Arena pits ~26 acceleration variants (LoRAs, full checkpoints, an API option and the 28-step baseline) against each other on pre-generated clips from a ~200-prompt corpus, via blind A/B voting: voters never see the variant names, just two clips, and pick the better one. Scores are Elo with confidence intervals; i2v and t2v are separate slices. (Sources: the Arena Space and arena/README.md.)

The tables below were fetched from the Arena's public API (/api/leaderboard; 26,593 i2v votes and 27,618 t2v votes at fetch time) on 2026-09-20 and re-ordered by Elo — no re-computation. Timing reference: RTX PRO 6000 Blackwell 96GB, bf16, no CPU offload (source: Arena timing_reference).

i2v slice · top 10

#VariantNFEElo ±CIVotesWin rateTimed speedupSource repo
1LightX2V Turbo 8-step v1.0 · native81086 ±63410542%3.07×lightx2v/Minimax-h3-Turbo
2larryvrh Turbo v4 · step600 EMA61084 ±62387145%4.12×larryvrh/MiniMax-H3-Turbo-Lora
3Plaguekind Parasyte Turbo61041 ±62414246%4.09×Plaguekind/H3-Lora
4MiniMax H3-Max (hosted via fal)—1035 ±70403244%—fal.ai
5silveroxides 4-to-8-step DARE merge61026 ±66416743%4.09×silveroxides/MiniMax-H3_tests
6VDN-H3 8-step (OpenVDN full ckpt)81010 ±5993839%3.17×OpenVDN/vdn-minimax-h3
7LightX2V Turbo 8-step v1.0 768p81007 ±56376038%3.07×lightx2v/Minimax-h3-Turbo
8LightX2V Turbo 4-step v0.141005 ±61299432%6.18×lightx2v/Minimax-h3-Turbo
9FastVideo FastH3 Preview v0.24999 ±56144436%—FastVideo/…-Preview-v0.2
10silveroxides DARE-TIES merge6999 ±56412344%4.12×silveroxides/MiniMax-H3_tests

Source: H3 Acceleration Arena public API, fetched 2026-09-28. “Timed speedup” is the Arena's own timing vs its 28-step baseline (not a WhichH3 estimate).

t2v slice · top 10

#VariantNFEElo ±CIVotesWin rateTimed speedupSource repo
1larryvrh Turbo v4 · step600 EMA61110 ±62330849%4.12×larryvrh/MiniMax-H3-Turbo-Lora
2MiniMax-H3 · 28-step baseline271073 ±66354347%1.0×MiniMax official weights
3silveroxides DARE-TIES merge61065 ±54317347%4.12×silveroxides/MiniMax-H3_tests
4Tutu AudioVideo 20→8 · ckpt10081042 ±67201839%3.09×tutututututu/Tutu-…-LoRA
5LightX2V Turbo 8-step v1.0 · native81029 ±63321144%3.07×lightx2v/Minimax-h3-Turbo
6VDN-H3 8-step (OpenVDN full ckpt)81029 ±71111445%3.17×OpenVDN/vdn-minimax-h3
7FastVideo FastH3 v1 VSA-DataFree · 6 steps (community config)61028 ±73414140%10.34×FastVideo/…-VSA-DataFree
8Plaguekind Parasyte Turbo61025 ±79314949%4.09×Plaguekind/H3-Lora
9alibaba-pai MiniMax-H3-Acc (PDD) 8-step81018 ±65309745%3.09×alibaba-pai/MiniMax-H3-Acc-LoRAs
10MiniMax H3-Max (hosted via fal)—1006 ±67321841%—fal.ai

Same source. Full 23 (i2v) / 27 (t2v) entries with confidence intervals: the Arena Space. Our archived copy: whichh3-data: acceleration_arena.json.

Three findings worth noting

  • larry v4 leads both: i2v #2 (1084) and t2v #1 (1110) — the only entry in the top three of both slices; the i2v crown goes to LightX2V 8-step (native, 1086), so the same family ranks very differently across slices (source: tables above).
  • Fast ≠ good-looking: the 28-step baseline still ranks #2 on t2v, while the 4-step family (6.18× timed speedup) sits at i2v #8 (source: tables above).
  • 6-NFE adapters remain the sweet spot: 3 of the i2v top 5 and 2 of the t2v top 3 are 6 NFE (~4.1× timed speedup) — nearly 4-step speed with near-baseline preference (source: tables above; Arena conventions: NFE = actual forwards, 6 NFE = 7 steps).

Mapping to this site (selectable in the calculator)

Arena entryi2v / t2v rankWhichH3 calculator entry
Plaguekind Parasyte Turbo#3 / #8Parasyte (6 NFE)
larryvrh Turbo v4 step600 EMA#2 / #1larry v4 8-step
alibaba-pai PDD 8-step#14 / #9PDD 8-step (official distill)
silveroxides DARE-TIES#10 / #3silveroxides (6 NFE)
LightX2V 8-step v1.0 (768p profile)#7 / #16LightX2V Turbo 8-step
LightX2V 4-step v1.1 768p#17 / #21LightX2V Turbo 4-step v1.1
FastVideo FastH3 (VSA)#12 / #7FastH3 4-step (VSA)
TaoMate-H3 3-step#19 / #24TaoMate 3-step

Pick any acceleration in step 2 of the calculator and its arena rank is shown right below (with the source). For stacking rules see the acceleration comparison; downloads live in the model library.

How to choose, by goal

  • Fast but not ugly (mostly i2v): the 6-NFE family — Parasyte / silveroxides DARE-TIES / larry v4 (~4.1× timed speedup, i2v top 5).
  • Safest, least fiddling: LightX2V 8-step (same as official turbo_mode) or PDD 8-step (official distill, needs its dedicated node — source: alibaba-pai).
  • t2v with quality first: silveroxides DARE-TIES (t2v #1), or expect less from aggressive step cuts.
  • Maximum speed: 4-step options (e.g. LightX2V 4-step) ≈6.2× timed speedup, but preference ranks clearly lower — spot-check before a batch.
  • Small VRAM / long clips: fix resolution and VRAM first (low-VRAM setups), then stack acceleration; attention backends (Sage/Kitchen/SLA/VSA) stack orthogonally.

Limitations (read before quoting)

  • It's community voting: every vote counts equally, one account can vote many times, and the sample is not everyone (source: Arena README, “Going public”).
  • The list mixes full checkpoints (OpenVDN, FastVideo) and an API option (fal H3-Max) with LoRAs — not the same usage model (source: Arena kind field).
  • Versions of the same family are separate entries (v1.0 / v1.1 / v0.2…); don't merge them (source: Arena variants.json).
  • Timed speedups only mean something on the same reference hardware (RTX PRO 6000 96GB, bf16, no offload); your absolute numbers will differ (source: Arena timing_reference).
  • WhichH3's calculator reports a resolution × duration matrix estimate (calibrated on 30+ community anchors) — a different measurement than the arena's timed runs (source: this site's data notes).
  • The leaderboard evolves; our snapshot is 2026-09-28 (votes still growing) — after that, the Arena is the source of truth.

Sources

Keep reading

H3 update alerts · at most weekly New LoRAs, new GPUs, leaderboard moves — one email

New acceleration methods, newly added GPUs, driver and ComfyUI gotchas. No update, no email. Unsubscribe any time; we only store your address.