MiniMax-H3 · fl2va × ref2va Hybrid b25-49 (W4A8, ~12.5 GB)

A 4-bit (asym_w4a8_int8) build of the community-standard b25-49 hybrid — the variant that, as far as I could find, did not exist in W4A8 form before this upload. Single file, drop-in for ComfyUI's stock Load Diffusion Model.

minimax_h3_hybrid_fl2va_ref2va_b25-49_w4a8_mixed.safetensors   12,540,858,008 bytes (11.68 GiB)   1132 tensors

What b25-49 means

MiniMax-H3 ships two checkpoints with identical architecture and tensor layout:

• fl2va — first/last-keyframe conditioning, noticeably better visual/audio quality, no reference slots.
• ref2va — adds image/video/audio reference conditioning, slightly lower base quality.

A tensor-by-tensor comparison shows the two are effectively identical everywhere except the per-block adaln_proj weights (the AdaLN modulation projections that route conditioning signal into each block). The bX-49 hybrids keep fl2va as the base and take adaln_proj from ref2va for blocks X–49 only.

b25-49 = blocks 25–49 from ref2va (the community default balance point). b30-49 leans quality, b20-49/b15-49 lean reference adherence.

How this file was built — no retraining, no re-quantization

This is a tensor transplant on already-quantized files, so no additional numerical re-quantization error is introduced:

The file header (143,408 B, including __metadata__._quantization_metadata) is byte-identical to the skeleton, so the quantization configuration is untouched and the data region keeps a 1:1 offset mapping.

Verification

• Non-quantized tensors: 332 / 332 byte-identical to the existing b25-49 INT8 hybrid — including all 100 per-block adaln_proj tensors, 200 norms, 17 token_refiner tensors and 15 top-level tensors.
• All 200 quantized linear layers dequantized per the official comfy-kitchen reference and compared against the INT8 hybrid: cosine 0.99725–0.99731 (mean 0.99730), relative RMS difference 7.33–7.42 % — the known size of the W4A8 quantization step (int8's own RMS error measures 0.88 %), i.e. the same weights at a coarser quantization, not a different model.
• Weight lineage spot-check: this file's linears sit 7.33–7.36 % from the fl2va-side hybrid and 7.82–8.49 % from a pure ref2va file — confirming the fl2va side was kept, as the bX-49 definition requires.
• Output correctness on-device: cos = 0.999948 against a dequantize-then-matmul reference.

Quantization profile

• 200 block linears (attn.qkv_proj, attn.out_proj, mlp.fc1, mlp.fc2 × 50 blocks): {"format": "asym_w4a8_int8", "group_size": 16, "convrot": true, "convrot_groupsize": 256}
• token_refiner, norms, adaln_proj, patch projections, condition_proj and output layers stay at BF16/F16/F32.
• Configuration is carried in __metadata__._quantization_metadata (the same convention as the source file), not in per-tensor comfy_quant blobs.

Hardware note (important)

The native W4A8 kernel path requires compute capability ≥ 8.0 (Ampere or newer) — AsymW4A8Int8Layout.MIN_SM_VERSION = (8, 0). On older cards the runtime falls back to the portable/eager implementation, which is correct but not faster:

So on Turing / GTX 16-series prefer the INT8 hybrid: it is ~2.6× faster per layer there, and this W4A8 file's only remaining benefit is halved file size. On sm_80+ this file is the better choice (smaller and faster). Note that quantization metadata is still detected fine on older cards — the format is not "disabled", the kernel simply runs through the fallback.

Usage (ComfyUI)

1. Put the file in ComfyUI/models/diffusion_models/.
2. Load with the stock Load Diffusion Model node (requires a ComfyUI version with INT8 ConvRot + W4A8 support; the log should show Detected mixed precision quantization).
3. The model routes as ref2va: keyframes via MiniMaxH3AddKeyframes, subject/scene/style references via MiniMaxH3ReferenceToVideo — both can be used in the same generation. Plain text-to-video and first-frame-only workflows work as usual.
4. Use the usual H3 text encoder and video/audio VAEs.

Credits & license

• Original weights: MiniMaxAI/MiniMax-H3 (FL2VA + REF2VA).
• ComfyUI packaging: Comfy-Org/MiniMax-H3.
• Quantized conversions used as source: Kijai/MiniMax-H3-experimental.
• Hybrid idea (adaln-only block splice): smhfacct/Minimax-H3-fl2va-ref2va-hybrid-models, also packaged by diobrando0.

No training, fine-tuning or gradient optimization was involved — this is tensor-level weight selection on top of existing quantized conversions. The merge/conversion is released under Apache-2.0; the underlying MiniMax H3 weights remain subject to the MiniMax H3 Community License. Please review it before use.

---

中文摘要

这是社区默认档 b25-49 混合模型的 W4A8 4bit 版本（11.68 GiB，1132 张量），此前社区没有这一档的 4bit 成品。 做法是在同一套量化文件之间做张量搬运，不重新量化：以 Kijai 的 fl2va_pruned_w4a8_mixed 为骨架， 只把 blocks.25–49 的 adaln_proj 覆盖为 ref2va 的值（50 个张量），header 与量化配置逐字节保持不变。

验证：332 个未量化张量与原 b25-49 INT8 版逐字节相同；200 个量化层反量化后余弦 0.9973、相对差 7.35%， 即 W4A8 量化步长本身的量级（int8 自身误差仅 0.88%），确认为同一模型的不同量化档。

硬件提示：W4A8 原生内核要求 sm_80+（Ampere 及以上）。在 sm_75（GTX 16 系 / Turing）上会走软件回退路径， 结果正确但更慢——同层实测比 INT8 慢约 2.6 倍。老卡请用 INT8 版；sm_80+ 的卡用这版既小又快。