Zhenzhen's AI Radar · MiniMax H3 · 2026-11-07 Evening Report
This issue features 12 new versions: Strongly Recommended 0 · Recommended 4 · Worth Watching 5 · Listed Only 3
Recommended (4)
【A】Alstroemeria Wren · v1.0 H3
LoRA · Port · Author MikoGoblin · Published 11-07 18:48 · Downloads 17
Created by Miko Goblin.
Image
Reasons for Recommendation
- Ported an older model to H3: A Krea 2 version existed previously
- Within 2 hours of release, it has 17 downloads and 1 likes, with a download rate higher than 87%% of H3 models from the last 7 days
- Includes 3 sample videos, allowing you to preview the results before deciding to download
- Documentation is relatively complete (approx. 3258 words)
- Free download, license allows in-site rental
How to Use
- Load this LoRA in the H3 workflow (148 MB · BF16), and include the trigger word "alstroemerialora" in the prompt
Trigger Word: alstroemerialora
【A】All In One Krea 2 Turbo Made Easy: ComfyUI Workflow for Krea 2 Turbo with WAN 2.2 or Minimax H3 or both or neither! · v1.0
Workflow · New Model · Author DXMPie · Published 11-07 15:23 · Downloads 31
One canvas for everything you do with Krea 2 Turbo. Turn it off and your words go through exactly as typed.
Image
Reasons for Recommendation
- Within 6 hours of release, it has 31 downloads and 1 likes, with a download rate higher than 82%% of H3 models from the last 7 days
- Includes 5 sample images, allowing you to preview the results before deciding to download
- The author provides specific parameters (3 instances), lowering the barrier to reproduction
- Free download, license allows in-site rental
How to Use
- ComfyUI Workflow: Krea2_Brain_Made_Easy_v1.0.zip (197 KB); after importing, follow the instructions to add custom nodes and models; this is not a new weight file
- Author's parameters: "Quick start (5 steps)"
- Author's parameters: "Krea 2 Turbo: 8 to 12 steps, CFG 1."
Risks and Limitations
- 15 sensitive-level examples have been hidden
- The author labels the base model as "Krea 2 Turbo"; judged as H3 content based on the name
- Dependent nodes and models need to be verified by yourself; this tool has not been run
【A】RefMod baker · v1.0
Workflow · New Model · Author Jas_A_Hook · Published 11-07 15:31 · Downloads 47
++ Help keep a character on-model” is accurate; it improves consistency without training a LoRA, but doesn’t guarantee an exact face match.++
Image
Reasons for Recommendation
- Within 6 hours of release, it has 47 downloads and 4 likes, with a download rate higher than 88%% of H3 models from the last 7 days
- Includes 3 sample images, allowing you to preview the results before deciding to download
- Documentation is relatively complete (approx. 953 words)
- Free download, license allows image generation, in-site rental, rental, sale, and sale of fusions
How to Use
- ComfyUI Workflow: Creating a refMOD.json (11 KB); after importing, follow the instructions to add custom nodes and models; this is not a new weight file
Risks and Limitations
- 1 sensitive-level examples have been hidden
- Dependent nodes and models need to be verified by yourself; this tool has not been run
【A】All in One MOSS-Image Made Easy: ComfyUI Workflow for MOSS-Image With Wan 2.2 or Minimax h3 or both or neither! · v1.0
Workflow · New Model · Author DXMPie · Published 11-07 17:03 · Downloads 22
The enhancer rewrites it into a full MOSS-style caption. Turn it off and your words go through exactly as typed.
Image
Reasons for Recommendation
- Within 4 hours of release, it has 22 downloads and 0 likes, with a download rate higher than 82%% of H3 models from the last 7 days
- Includes 3 sample images, allowing you to preview the results before deciding to download
- The author provides specific parameters (3 instances), lowering the barrier to reproduction
- Free download, license allows in-site rental and image generation
How to Use
- ComfyUI Workflow: MOSS_Brain_Made_Easy_v1.0.zip (129 KB); after importing, follow the instructions to add custom nodes and models; this is not a new weight file
- Author's parameters: "## Quick start (4 steps)"
- Author's parameters: "Normal SFT or Base at about 30 steps, CFG 3, euler (the official guide suggests 50 or mor…"
Risks and Limitations
- 11 sensitive-level examples have been hidden
- The author labels the base model as "MOSS Image"; judged as H3 content based on the name
- Dependent nodes and models need to be verified by yourself; this tool has not been run
Worth Watching (5)
【B】Modern Beauty - MMH3 · v1.0
LoRA · New Model · Author DuckyDuo · Published 11-07 11:23 · Downloads 210
No trigger word required Intended to bring MMH3's beauty aesthetics to a more modern standard, diversifying the datatset is the plan for v2 eventually.
Image
Reasons for Recommendation
- Within 10 hours of release, it has 210 downloads and 10 likes, with a download rate higher than 97%% of H3 models from the last 7 days
- Includes 1 sample videos + 1 sample images, allowing you to preview the results before deciding to download
- Free download, license allows in-site rental
How to Use
- Load this LoRA in the H3 workflow (142 MB · BF16); the author has not registered a trigger word, refer to the example prompts to describe the effect
Risks and Limitations
- 6 sensitive-level examples have been hidden
- No trigger word registered
【B】Mooshie · v1.0 H3
LoRA · Port · Author MikoGoblin · Published 11-07 20:01 · Downloads 6
Mooshie rarely feels any urgency to fill a conversation simply because nobody has spoken for thirty seconds, which makes her company unusually restful for peop…
Image
Reasons for Recommendation
- Ported an older model to H3: Previous versions existed for Krea 2 / LTXV 2.3 / ZImageTurbo
- Within 1 hours of release: 6 downloads, 1 likes
- Documentation is relatively complete (approx. 2652 words)
- Free download, license allows in-site rental
How to Use
- Load this LoRA in the H3 workflow (148 MB · BF16), and include the trigger word "mooshielora" in the prompt
Risks and Limitations
- All examples are sensitive-level (2 hidden), making it impossible to preview the effect
- The card image is an AI-generated illustration, not the actual model output; the author has no publicly available general-level examples
Trigger Word: mooshielora
【B】Maria Von Vergin · MiniMaxH3
LoRA · Port · Author MisterMR · Published 11-07 19:27 · Downloads 24
This LoRA does not depict a real person;
Image
Reasons for Recommendation
- Ported an older model to H3: Previous versions existed for Pony / SD 1.5 / SDXL 1.0
- Within 2 hours of release, it has 24 downloads and 2 likes, with a download rate higher than 94%% of H3 models from the last 7 days
- Documentation is relatively complete (approx. 1652 words)
- Free download, license allows in-site rental
How to Use
- Load this LoRA in the H3 workflow (296 MB); the author has not registered a trigger word, refer to the example prompts to describe the effect
Risks and Limitations
- All examples are sensitive-level (18 hidden), making it impossible to preview the effect
- No trigger word registered
- The card image is an AI-generated illustration, not the actual model output; the author has no publicly available general-level examples
【B】Heather Mason - Silent Hill 3 · H3 Heather
LoRA · New Model · Author GayLizardSpy · Published 11-07 16:59 · Downloads 30
It was trained on cutscenes from the game and usually produces a 3d style.
Image
Reasons for Recommendation
- Within 4 hours of release, it has 30 downloads and 5 likes, with a download rate higher than 87%% of H3 models from the last 7 days
- Free download, license allows in-site rental
How to Use
- Load the LoRA in the H3 workflow (148 MB · BF16); include the trigger word "sh3heather" in the prompt; the author also provides a complete example prompt, see the copyable tag below
Risks and Limitations
- All examples are sensitive-level (4 hidden), so effects cannot be previewed
- The card image is an AI-generated schematic, not the actual model output; the author has no publicly available non-sensitive examples
Trigger Word: sh3heather, a young woman with short blonde hair, wearing a white vest over an orange sleeveless turtleneck, a short green skirt, and brown boots.
[B] Annabelle NordhavenLove · MiniMaxH3
LoRA · Port · Author MisterMR · Published 11-07 15:05 · Downloads 33
This LoRA does not depict a real person; The Nordhaven family represented a rare, almost legendary presence along the rugged coast of Bohuslän.
Image
Reasons for Recommendation
- Ported from an older model to H3: Previous versions existed for LTXV 2.5 / Krea 2 / Pony
- Within 6 hours of release, it has 33 downloads and 2 likes, with a download rate higher than 83%% of H3 models over the last 7 days
- The description is relatively complete (approx. 2867 words)
- Free download; the license permits in-site rental, leasing, and image generation
How to Use
- Load the LoRA in the H3 workflow (296 MB); the author has not registered a trigger word, so refer to the example prompt to describe the effect
Risks and Limitations
- All examples are sensitive-level (17 hidden), so effects cannot be previewed
- No trigger word registered
- The card image is an AI-generated schematic, not the actual model output; the author has no publicly available non-sensitive examples
Included Only (3)
- Aerith Gainsborough Voice (Eng) - Final Fantasy 7 - Audio Only [Minimax H3] (Commission) · Minimax H3 v1.0 — LoRA: This Lora was trained on English language audio clips of Aerith from the FF7 Remake trilogy. The dataset was captioned & transcribed in English.
- KinetiX Dynamic Flash Step Slider · v1.0 — LoRA: Visual Effect: Linearly scales character spatial displacement from realistic continuous foot strides to instantaneous supersonic quantum flash-steps with optic…
- Zaynab NordhavenLove · MiniMaxH3 — LoRA: This LoRA does not depict a real person;
Data: Civitai public API (base model MiniMax H3), inclusion window 11-07 08:31 → 11-07 21:01 (Beijing Time). Reasons for recommendation are automatic drafts; download counts only reflect attention.