docs: add v0.34.3 changelog and CMS release notes

.cursor/skills/cms-changelog-sync/SKILL.md

@@ -3,9 +3,11 @@ name: cms-changelog-sync
 description: >-
   Sync ComfyUI release notes to Strapi CMS: LLM-simplify English changelog for
   in-app popup, translate to zh/ja/ko/fr/ru/es in staging, push drafts to CMS.
-  Use when updating changelog/index.mdx for CMS, running cms:prepare/cms:sync,
-  Strapi release-notes, published-versions.json, CMS staging, simplifying
-  release notes for the notification popup, or cms:publish to go live.
+  Resolves docs/local/cloud bullet URLs (blog.comfy.org, workflow_templates
+  index.json, Cloud ?template=, user UTM, GitHub PRs). Use when updating
+  changelog/index.mdx for CMS, running cms:prepare/cms:sync, Strapi
+  release-notes, published-versions.json, CMS staging, simplifying release
+  notes for the notification popup, or cms:publish to go live.
 ---
 
 # CMS Changelog Sync
@@ -101,6 +103,32 @@ Config: `.github/scripts/cms/cms-config.json` → `simplify`
 
 Style: principle-only prompt in `cms-simplify-prompt.ts` (no concrete version examples — avoids LLM contamination).
 
+`prepare:en` copies docs URLs and local-length copy into Cloud. After it runs, rewrite **Cloud EN** (links and wording) before translating. Never invent URLs.
+
+**Copy length (local vs Cloud):** Cloud popup users skim. After merge, shorten Cloud bullets so they do not list every node, mode, or task type. One short clause is enough: added the model, or one capability. Local CMS (`staging/en/`) and docs `changelog/index.mdx` can keep the fuller scope (which nodes, which modes). Do not shorten local to match Cloud.
+
+Example: docs/local may say H3 Max landed on text-to-video, first-last-frame, and reference nodes. Cloud: `Added H3 Max model support`.
+
+## Bullet links (docs, local CMS, Cloud CMS)
+
+Resolve each feature bullet **before** `cms:prepare:locales`. Search these sources every time a new version lands:
+
+| Source | Where |
+|--------|--------|
+| Template index | [templates/index.json](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/index.json) (raw: `https://raw.githubusercontent.com/Comfy-Org/workflow_templates/main/templates/index.json`) |
+| Blog | [blog.comfy.org](https://blog.comfy.org/) ([archive](https://blog.comfy.org/archive)) |
+
+Match a template by `name`, `title`, or `models` to the changelog item. Match a blog post only if it covers **this** product or version (MiniMax H3 day-0 is not MiniMax H3 Max).
+
+**Video templates:** when several templates exist, pick one by suffix on `name`, in this order: **r2v → i2v → t2v**. Example: `api_minimax_h3_max_r2v` over `_i2v` / `_t2v`. If none of those suffixes exist, use the remaining matching template (`flf2v`, `edit`, and similar). Cloud URL shape: `https://cloud.comfy.org/?template=<name>` (no UTM unless the user supplied one).
+
+| Surface | Link priority (first match wins) |
+|---------|----------------------------------|
+| **Cloud** CMS (`staging/cloud/`) | 1. UTM / `links.comfy.org` URL **the user provided** 2. Cloud template URL from the index 3. GitHub PR 4. ComfyUI repo commit/tag/compare |
+| **Docs** `changelog/index.mdx` and **local** CMS (`staging/en/` etc.) | 1. Matching [blog.comfy.org](https://blog.comfy.org/) post 2. GitHub PR 3. ComfyUI repo commit/tag/compare |
+
+Do **not** put Cloud `?template=` URLs on docs or local popup. Do **not** copy local blog/PR links onto Cloud when a template (or user UTM) exists. After `prepare:en` merges comfyui → cloud, replace Cloud bullets that still point at PRs if the index has a template.
+
 ## ComfyUI-WIKI commits (omit from changelog)
 
 When curating `changelog/index.mdx` from ComfyUI git history, **do not add bullets** for commits routinely opened by **[ComfyUI-WIKI](https://github.com/Comfy-Org/ComfyUI-WIKI)** — they are dependency/content syncs, not core release features:
@@ -185,15 +213,15 @@ Requires **Bun**. Loads `.env.local` automatically.
 
 ### New release version
 
-1. Add full `<Update>` block to `changelog/index.mdx` (docs quality — unchanged).
+1. Add full `<Update>` block to `changelog/index.mdx` (docs quality — unchanged). Set each bullet URL using **Bullet links** (blog → PR → repo for docs).
 
 2. **Step 1 — Simplify EN** — review before translating:
 
    ```bash
    pnpm cms:prepare:en -- --force v0.25.1
    ```
 
-   Inspect: `.github/scripts/cms/staging/en/changelog/index.mdx` → **stop until approved**
+   Inspect: `.github/scripts/cms/staging/en/changelog/index.mdx` (blog/PR/repo, fuller copy). Rewrite `.github/scripts/cms/staging/cloud/en/changelog/index.mdx`: user UTM or `?template=` from the index, and **shorter** bullets (model support, not every node). → **stop until approved**
 
 3. **Step 2 — Translate** — from approved staging EN only:
 
@@ -265,8 +293,10 @@ Re-run with `--force`. Staging without `--force` **skips** existing `<Update>` b
 When user asks to update CMS release notes:
 
 - [ ] Confirm `changelog/index.mdx` has the new `<Update>` block
+- [ ] Resolve bullet URLs: search template `index.json` and [blog.comfy.org/archive](https://blog.comfy.org/archive); Cloud = user UTM then `?template=` (video r2v → i2v → t2v); docs/local = blog then PR then repo
+- [ ] Shorten Cloud EN bullets (added model support, skip node lists). Keep local/docs more detailed
 - [ ] Omit ComfyUI-WIKI items (embedded docs, workflow templates, model blueprints) unless user explicitly asks
-- [ ] Run `pnpm cms:prepare:en`; show staging EN → **wait for user approval**
+- [ ] Run `pnpm cms:prepare:en`; rewrite Cloud EN links; show staging EN → **wait for user approval**
 - [ ] Run `pnpm cms:prepare:locales` (not `cms:prepare:en`) → **wait for user approval**
 - [ ] Run `pnpm cms:preview -- --project comfyui ...` then `pnpm cms:sync -- --project comfyui ...` **only after user confirms staging**
 - [ ] Run cloud `cms:sync` / `cms:publish` only after separate explicit cloud confirmation

.cursor/skills/docs-i18n-translate/SKILL.md

@@ -104,6 +104,8 @@ only re-translate changed `##` sections when `auto_chunk` applies).
 
 Do not add bullets for dependency-only version bumps. See also **`cms-changelog-sync`** for CMS popup rules.
 
+**Docs changelog bullet URLs** (same as local CMS): matching [blog.comfy.org](https://blog.comfy.org/) post first, then the GitHub PR, then the ComfyUI repo commit/tag/compare. Do not use Cloud `?template=` links on the docs changelog. Cloud popup URLs are a separate rule in **cms-changelog-sync**.
+
 ```bash
 pnpm translate -- changelog/index.mdx
 pnpm translate -- changelog/index.mdx --lang zh

.github/scripts/cms/README.md

@@ -62,6 +62,8 @@ Each version is **saved immediately** after simplify/translate (safe to resume).
 
 After prepare:en, secondary projects (cloud) merge only the prepared `<Update>` blocks from comfyui. If cloud EN already has that version, incoming tracking shortlinks are rewritten to match cloud EN (local vs cloud campaign URLs). `prepare:locales` translates each project from its own staging EN and does not copy comfyui locale files onto cloud.
 
+**Bullet URLs** (agents rewrite after prepare:en; see skill `cms-changelog-sync`): Cloud prefers a user-supplied UTM/`links.comfy.org` URL, else `https://cloud.comfy.org/?template=<name>` from [workflow_templates `index.json`](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/index.json) (video: r2v → i2v → t2v). Docs and local CMS prefer a [blog.comfy.org](https://blog.comfy.org/) post, then the GitHub PR, then the repo commit/tag. **Cloud copy** is shorter than local (announce the model, do not enumerate every node).
+
 ## Projects & attention
 
 | Project | CLI | Default attention |

.github/scripts/cms/published-versions.json

@@ -776,7 +776,7 @@
         "ru",
         "zh"
       ],
-      "published_at": "2026-08-28T15:40:36.700Z"
+      "published_at": "2026-08-28T15:54:39.895Z"
     },
     {
       "project": "comfyui",
@@ -790,7 +790,35 @@
         "ru",
         "zh"
       ],
-      "published_at": "2026-08-28T15:39:47.293Z"
+      "published_at": "2026-08-28T15:54:14.650Z"
+    },
+    {
+      "project": "cloud",
+      "version": "0.34.3",
+      "locales": [
+        "en",
+        "es",
+        "fr",
+        "ja",
+        "ko",
+        "ru",
+        "zh"
+      ],
+      "published_at": "2026-09-03T10:13:11.924Z"
+    },
+    {
+      "project": "comfyui",
+      "version": "0.34.3",
+      "locales": [
+        "en",
+        "es",
+        "fr",
+        "ja",
+        "ko",
+        "ru",
+        "zh"
+      ],
+      "published_at": "2026-09-03T10:12:48.455Z"
     }
   ]
 }

.github/scripts/cms/staging/cloud/en/changelog/index.mdx

@@ -4,23 +4,30 @@
 ---
 
 
+<Update label="v0.34.3" description="September 2, 2026">
+
+**Partner Node Updates**
+* [**MiniMax H3 Max**](https://cloud.comfy.org/?template=api_minimax_h3_max_r2v): Added H3 Max model support
+
+</Update>
+
 <Update label="v0.34.2" description="August 27, 2026">
 
 **Partner Node Updates**
 * [**Gemini Omni 1.1 Flash**](https://links.comfy.org/4qIN5DH): Faster video generation and conversational editing, with 4K output and video extend
 
 </Update>
 
 <Update label="v0.34.0" description="August 26, 2026">
 
 **New Open-Source Model Support**
 * [**Pixal3D & TRELLIS2**](https://cloud.comfy.org/?template=3d_pixal3d_trellis2_image_to_model): 3D mesh and texture generation with remesh, decimate, UV unwrap, texture bake
 * [**SAM 3D Body**](https://cloud.comfy.org/?template=utility_sam3d_body): Dependency-free 3D body detection and segmentation with pose rendering and animation export
 * [**TAESD H3**](https://github.com/Comfy-Org/ComfyUI/pull/15695): Lightweight VAE decoder for faster previews
 
 **Partner Node Updates**
 * [**PixVerse V6**](https://github.com/Comfy-Org/ComfyUI/pull/15880): Added V6 text-to-video, image-to-video, first-last-frame, extend, fusion nodes with native audio
 * [**Tripo**](https://github.com/Comfy-Org/ComfyUI/pull/15851): Removed the retired Refine Draft model node and v2.0 model version
 
 </Update>
 
@@ -29,8 +36,8 @@
 **Partner Node Updates**
 * [**Wan 3.0**](https://cloud.comfy.org/?template=api_wan3_0_i2v): Text, image, and reference-to-video generation with native audio up to 30 seconds
 * [**Meshy-7**](https://cloud.comfy.org/?template=api_meshy7_image_to_model): Adds meshy-7 and ultra mode to text, image, and multi-image-to-3D nodes
 * [**ByteDance vCube Video Enhance**](https://cloud.comfy.org/?template=api_bytedance_vcube_video_enhance): Upscales and restores video up to 8K with optional frame interpolation
 * [**Kling v2**](https://github.com/Comfy-Org/ComfyUI/pull/15676): Removes kling-v2 from Kling Image Generation ahead of September 15 retirement
 
 </Update>
 
@@ -38,9 +45,9 @@
 
 **Partner Node Updates**
 * [**Flux Video Upscale**](https://github.com/Comfy-Org/ComfyUI/pull/15753): Upscale video 1.5x to 3x with FLUX super-resolution
 * [**ByteDance Seedream Fast Mode**](https://github.com/Comfy-Org/ComfyUI/pull/15750): Fast prompt-optimization mode for Seedream 5.0 Pro image-to-image
 * [**Gemini 3.7 Flash**](https://github.com/Comfy-Org/ComfyUI/pull/15688): Gemini 3.7 Flash added to the Gemini text node
 * [**Ideogram & Pruna P-Image**](https://github.com/Comfy-Org/ComfyUI/pull/15751): Renamed Ideogram P-Image to credit Pruna
 
 </Update>
 
@@ -48,8 +55,8 @@
 
 **Partner Node Updates**
 * [**Fish Audio**](https://github.com/Comfy-Org/ComfyUI/pull/15612): Adds text-to-speech, speech-to-text, voice selector, and instant voice clone nodes
 * [**ByteDance Seedance 2.5 1080p**](https://github.com/Comfy-Org/ComfyUI/pull/15684): Adds 1080p resolution support for Seedance 2.5
 * [**ByteDance Seedance 2.5 Extend**](https://github.com/Comfy-Org/ComfyUI/pull/15579): Adds extend task type for video continuation
 
 </Update>
 
@@ -61,7 +68,7 @@
 
 **Partner Node Updates**
 * [**MiniMax H3 Context IR & Regenerate**](https://github.com/Comfy-Org/ComfyUI/pull/15471): Added Context IR prompt enhancer and Regenerate to 2K nodes
 * [**Bria GenFill, Eraser, Expand & Upscale**](https://github.com/Comfy-Org/ComfyUI/pull/15572): Added Generative Fill, Eraser, Expand Image, and Increase Resolution nodes
 
 </Update>
 
@@ -71,7 +78,7 @@
 * [**LTX 2.5**](https://links.comfy.org/cloLTX2_5): Native support with STG, dual CFG, duration prediction
 
 **Partner Node Updates**
 * [**Qwen Image 3.0**](https://links.comfy.org/45W0b6L): Adds 3.0 and 3.0 Pro text-to-image and edit nodes
 * [**LTX 2.5**](https://links.comfy.org/3TV1HDB): Text, image, and audio to video partner nodes
 * [**Grok Imagine Image 2.0**](https://github.com/Comfy-Org/ComfyUI/pull/15496): Adds support for the grok-imagine-image-2.0 model
 
@@ -84,8 +91,8 @@
 
 **Partner Node Updates**
 * [**ByteDance SeeDance 2.5**](https://links.comfy.org/3RPYzbv): Added ByteDance SeeDance 2.5 model support
 * [**Seedream 5.0 Layer Separation**](https://links.comfy.org/4hgRB9L): ByteDance Seedream 5.0 Pro Layer Separation Support
 * [**Reve deprecated**](https://github.com/Comfy-Org/ComfyUI/pull/15331): Reve partner nodes have been deprecated
 
 **New Node Updates**
 * [**Create Layered Image**](https://github.com/Comfy-Org/ComfyUI/pull/15317): PSD-style layer compositing with blend modes and layer state
@@ -101,7 +108,7 @@
 **Partner Node Updates**
 * [**BFL Flux 3 Video**](https://links.comfy.org/3TVdfGP): Added Flux 3 video model support
 * [**TopazAI Bloom 2 & Wonder 3.5**](https://github.com/Comfy-Org/ComfyUI/pull/15294): Added Bloom 2 and Wonder 3.5 model support
 * [**Kling EOL**](https://github.com/Comfy-Org/ComfyUI/pull/15249): Removed retired legacy Kling models and Virtual Try-On API
 
 </Update>
 
@@ -116,12 +123,12 @@
 
 **New Open-Source Model Support**
 * [**MiniMax-H3**](https://links.comfy.org/3TuT9TO): Adds audio-video model support for MiniMax-H3
 * [**Pruna LTX VAE**](https://github.com/Comfy-Org/ComfyUI/pull/15129): PrunaVAED support for faster LTX 2.3 decoding
 
 **New Node Updates**
 * [**Save Video CRF**](https://github.com/Comfy-Org/ComfyUI/pull/15191): Adds CRF option to Save Video node
 * [**SaveText CSV**](https://github.com/Comfy-Org/ComfyUI/pull/15217): SaveText node can now save .csv output
 * [**VAEDecodeAudio**](https://github.com/Comfy-Org/ComfyUI/pull/15211): Handles decoding of nested audio latents
 
 **Partner Node Updates**
 * [**xAI Grok Imagine Video 1.5**](https://github.com/Comfy-Org/ComfyUI/pull/15197): Updated xAI nodes for grok-imagine-video-1.5 support
@@ -144,13 +151,13 @@
 * [**Gemma4 12B**](https://github.com/Comfy-Org/ComfyUI/pull/14304): Gemma4 12B model support
 * [**MageFlow**](https://github.com/Comfy-Org/ComfyUI/pull/15026): MageFlow model support
 * [**Uni3C ControlNet**](https://github.com/Comfy-Org/ComfyUI/pull/14946): Native Uni3C ControlNet support for Wan models
 * [**Anima**](https://github.com/Comfy-Org/ComfyUI/pull/14954): Anima llite control model support
 
 **New Node Updates**
 * [**Trainer video processing**](https://github.com/Comfy-Org/ComfyUI/pull/13588): Video processing nodes and Image Processing Node video support
 
 **Partner Node Updates**
 * [**Recraft V4.1**](https://github.com/Comfy-Org/ComfyUI/pull/15105): Recraft V4.1 models and V4 image edit nodes
 * [**ByteDance audio multilingual**](https://github.com/Comfy-Org/ComfyUI/pull/15034): ByteDance audio-1.0-multilingual model support
 * [**OpenRouter**](https://github.com/Comfy-Org/ComfyUI/pull/15021): New OpenRouter models, including Claude Opus 5
 * [**Anthropic**](https://github.com/Comfy-Org/ComfyUI/pull/15023): New Anthropic models, including Claude Opus 5
@@ -160,7 +167,7 @@
 <Update label="v0.28.3" description="July 22, 2026">
 
 **Partner Node Updates**
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/15014): Pass videos as inline data in Gemini Omni partner node
 * [**OpenRouter**](https://github.com/Comfy-Org/ComfyUI/pull/15021): New OpenRouter models
 * [**Anthropic**](https://github.com/Comfy-Org/ComfyUI/pull/15023): New Anthropic models
 
@@ -169,7 +176,7 @@
 <Update label="v0.28.2" description="July 18, 2026">
 
 **Partner Node Updates**
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/14984): Switched Omni model to the Interactions API
 
 </Update>
 
@@ -185,19 +192,19 @@
 <Update label="v0.28.0" description="July 15, 2026">
 
 **New Open-Source Model Support**
 * [**SeedVR2**](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/utility_seedvr2_3b_int8_upscale_image.json): Native SeedVR2 image and video upscaling
 * [**PID 1.5**](https://github.com/Comfy-Org/ComfyUI/pull/14894): Added PixelDiT PID 1.5 model support
 * [**int4 Models**](https://github.com/Comfy-Org/ComfyUI/pull/14859): convrot int4 support and Turing int8/int4 optimizations
 
 **New Node Updates**
 * [**Save 3D (Advanced)**](https://github.com/Comfy-Org/ComfyUI/pull/14701): Save 3D (Advanced), Save Splat, and Save Point Cloud nodes
 * [**SaveText**](https://github.com/Comfy-Org/ComfyUI/pull/14102): Save text outputs to .txt, .md, or .json
 * [**Text Overlay**](https://github.com/Comfy-Org/ComfyUI/pull/14610): Text overlay for images and video
 * [**Text**](https://github.com/Comfy-Org/ComfyUI/pull/14870): Built-in Text node re-enabled after undeprecation
 
 **Partner Node Updates**
 * [**sync.so sync-3**](https://github.com/Comfy-Org/ComfyUI/pull/14928): sync.so sync-3 lip-sync and talking-image support
 * [**Seedream thinking control**](https://github.com/Comfy-Org/ComfyUI/pull/14853): Option to disable thinking on Seedream nodes
 * [**Gemini Image**](https://github.com/Comfy-Org/ComfyUI/pull/14917): Gemini Image preview models moved to release versions
 
 </Update>
@@ -205,7 +212,7 @@
 <Update label="v0.27.1" description="July 8, 2026">
 
 **Partner Node Updates**
 * [**Seedream 5.0 Pro**](https://links.comfy.org/4ydIJYi): ByteDance image generation model support added.
 * [**Seed Audio 1.0**](https://github.com/Comfy-Org/ComfyUI/pull/14731): Speech, music, sound effects, multi-speaker dialogue.
 * [**Ideogram V1/V2 deprecated**](https://github.com/Comfy-Org/ComfyUI/pull/14712): Legacy Ideogram partner nodes deprecated.
 * [**StabilityAI temporarily removed**](https://github.com/Comfy-Org/ComfyUI/pull/14737): StabilityAI partner nodes temporarily removed.
@@ -215,15 +222,15 @@
 <Update label="v0.27.0" description="June 30, 2026">
 
 **Int8 Support**
 * [**int8 convrot models**](https://github.com/Comfy-Org/ComfyUI/pull/14636): Support for int8 quantized convrot models with faster inference on Turing GPUs
 
 **Partner Node Updates**
 * [**Nano Banana 2 Lite**](https://github.com/Comfy-Org/ComfyUI/pull/14693): Google's lightweight image generation model
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/14695): Video generation node powered by Gemini
 
 **New Node Updates**
 * [**Seed Node**](https://github.com/Comfy-Org/ComfyUI/pull/14616): Simplifies seed management for reproducible workflows
 * [**Advanced Krea 2 Model Merging**](https://github.com/Comfy-Org/ComfyUI/pull/14621): Merges Krea 2 models
 * [**Bounding Box Canvas**](https://github.com/Comfy-Org/ComfyUI/pull/14537): Canvas with Ideogram JSON prompt support
 * [**ConditioningMultiply**](https://github.com/Comfy-Org/ComfyUI/pull/14686): Multiplies conditioning values
 * [**GLSL Update**](https://github.com/Comfy-Org/ComfyUI/pull/13195): GLSL node updated to use ANGLE library
@@ -241,7 +248,7 @@
 
 **Partner Node Updates**
 * [**SeeDance 2.0**](https://github.com/Comfy-Org/ComfyUI/pull/14614): Added 4K resolution support for ByteDance SeeDance 2.0
 * [**HappyHorse 1.1**](https://links.comfy.org/3SoBFYI): Added support for Alibaba's HappyHorse 1.1 model
 * [**Grok Video**](https://github.com/Comfy-Org/ComfyUI/pull/14612): Added 1080p support for Grok video generation
 
 </Update>
@@ -249,24 +256,24 @@
 <Update label="v0.26.0" description="June 23, 2026">
 
 **New Open-Source Model Support**
 * [**Krea2**](https://links.comfy.org/4eC8714): Krea's open-source image model with wide aesthetic and stylistic range
 * [**Boogu-Image**](https://github.com/Comfy-Org/ComfyUI/pull/14523): Text-to-image with Qwen3-VL-8B encoder and image editing
 * [**Qwen3-VL Text Generation**](https://github.com/Comfy-Org/ComfyUI/pull/14298): Image-aware text generation with 4B and 8B models
 * [**SCAIL-2 Multireference**](https://github.com/Comfy-Org/ComfyUI/pull/14509): Multi-reference SCAIL-2 character replacement
 
 **New Node Updates**
 * [**LTX2 Context Windows**](https://github.com/Comfy-Org/ComfyUI/pull/13325): Context window sampling for LTX-2 video+audio with IC-LoRA splicing
 * [**Save Nodes Output Socket**](https://github.com/Comfy-Org/ComfyUI/pull/13866): Output socket on save nodes for downstream wiring
 
 **Partner Node Updates**
 * [**Luma Rays 3.2**](https://github.com/Comfy-Org/ComfyUI/pull/14540): Seven nodes for Luma Ray 3.2 video generation
 
 </Update>
 
 <Update label="v0.25.1" description="June 16, 2026">
 
 **Partner Node Updates**
 * [**Kling V3-Turbo**](https://github.com/Comfy-Org/ComfyUI/pull/14528): Text-to-video generation model
 
 </Update>
 
@@ -286,9 +293,9 @@
 <Update label="v0.24.1" description="June 3, 2026">
 
 **Partner Node Updates**
 * **Bria Green Background**: Image green background removal node.
 * **Bria Transparent Video Background**: Video background removal with alpha channel output.
 * **Krea 2 Medium Turbo**: Faster medium-sized image generation model.
 * **Seed input to Flux Erase node**: Adds seed parameter for repeatable erasure results.
 
 </Update>
@@ -299,7 +306,7 @@
 * [**Microsoft Lens**](https://github.com/Comfy-Org/ComfyUI/pull/14077): Text-to-image model with GPT-OSS-20B encoder
 * [**NVIDIA PixelDiT & PiD**](https://github.com/Comfy-Org/ComfyUI/pull/14103): Text-to-image models including PixelDiT 1300M and PiD family
 * [**TripoSplat**](https://github.com/Comfy-Org/ComfyUI/pull/14210): Image-to-3D Gaussian splat model
 * [**MediaPipe Face Detection**](https://github.com/Comfy-Org/ComfyUI/pull/14009): Dependency-free face detection and landmarker
 
 Partner Node Updates
 * **OpenRouter LLM Node**: Partner node for OpenRouter language model API
@@ -310,7 +317,7 @@
 
 **Partner Node Updates**
 * **Krea2 Image nodes**: Added new image generation nodes for Krea2 workflows
 * **SeeDance 2**: Improved video reference uploading via memoryview
 
 </Update>
 
@@ -326,7 +333,7 @@
 
 * **OpenRouter LLM node**: New partner node for OpenRouter API access.
 * **Anthropic node**: Added reasoning widget for step-by-step thought display.
 * **ByteDance2Reference node**: Added automatic upscaling widget for video outputs.
 * **Grok LLM**: Fixed issue passing images to the model.
 
 </Update>

.github/scripts/cms/staging/cloud/es/changelog/index.mdx

@@ -4,329 +4,336 @@
 ---
 
 
+<Update label="v0.34.3" description="2 de septiembre de 2026">
+
+**Actualizaciones de nodos de socios**
+* [**MiniMax H3 Max**](https://cloud.comfy.org/?template=api_minimax_h3_max_r2v): Se ha añadido compatibilidad con el modelo H3 Max
+
+</Update>
+
 <Update label="v0.34.2" description="27 de agosto de 2026">
 
 **Actualizaciones de nodos de socios**
 * [**Gemini Omni 1.1 Flash**](https://links.comfy.org/4qIN5DH): Generación de video más rápida y edición conversacional, con salida 4K y extensión de video
 
 </Update>
 
 <Update label="v0.34.0" description="26 de agosto de 2026">
 
 **Nuevo soporte para modelos de código abierto**
 * [**Pixal3D y TRELLIS2**](https://cloud.comfy.org/?template=3d_pixal3d_trellis2_image_to_model): Generación de mallas y texturas 3D con remallado, decimación, desarrollo UV y horneado de texturas
 * [**SAM 3D Body**](https://cloud.comfy.org/?template=utility_sam3d_body): Detección y segmentación de cuerpo 3D sin dependencias, con renderizado de pose y exportación de animaciones
 * [**TAESD H3**](https://github.com/Comfy-Org/ComfyUI/pull/15695): Decodificador VAE ligero para vistas previas más rápidas
 
 **Actualizaciones de nodos de socios**
 * [**PixVerse V6**](https://github.com/Comfy-Org/ComfyUI/pull/15880): Se añadieron nodos V6 de texto a video, imagen a video, primer-último fotograma, extensión y fusión con audio nativo
 * [**Tripo**](https://github.com/Comfy-Org/ComfyUI/pull/15851): Se eliminaron el nodo de modelo Refine Draft retirado y la versión de modelo v2.0
 
 </Update>
 
 <Update label="v0.33.4" description="24 de agosto de 2026">
 
 **Actualizaciones de nodos asociados**
 * [**Wan 3.0**](https://cloud.comfy.org/?template=api_wan3_0_i2v): Generación de texto, imagen y referencia a video con audio nativo de hasta 30 segundos
 * [**Meshy-7**](https://cloud.comfy.org/?template=api_meshy7_image_to_model): Añade meshy-7 y modo ultra a los nodos de texto, imagen y multi-imagen a 3D
 * [**ByteDance vCube Video Enhance**](https://cloud.comfy.org/?template=api_bytedance_vcube_video_enhance): Aumenta la resolución y restaura video hasta 8K con interpolación de fotogramas opcional
 * [**Kling v2**](https://github.com/Comfy-Org/ComfyUI/pull/15676): Elimina kling-v2 de Kling Image Generation antes de su retiro el 15 de septiembre
 
 </Update>
 
 <Update label="v0.33.3" description="20 de agosto de 2026">
 
 **Actualizaciones de nodos de socios**
 * [**Flux Video Upscale**](https://github.com/Comfy-Org/ComfyUI/pull/15753): Amplía videos de 1.5x a 3x con superresolución FLUX
 * [**ByteDance Seedream Fast Mode**](https://github.com/Comfy-Org/ComfyUI/pull/15750): Modo rápido de optimización de prompts para image-to-image con Seedream 5.0 Pro
 * [**Gemini 3.7 Flash**](https://github.com/Comfy-Org/ComfyUI/pull/15688): Gemini 3.7 Flash añadido al nodo de texto Gemini
 * [**Ideogram & Pruna P-Image**](https://github.com/Comfy-Org/ComfyUI/pull/15751): Renombrado Ideogram P-Image para acreditar a Pruna
 
 </Update>
 
 <Update label="v0.33.2" description="17 de agosto de 2026">
 
 **Actualizaciones de nodos de socios**
 * [**Fish Audio**](https://github.com/Comfy-Org/ComfyUI/pull/15612): Añade nodos de texto a voz, voz a texto, selector de voz y clonación instantánea de voz
 * [**ByteDance Seedance 2.5 1080p**](https://github.com/Comfy-Org/ComfyUI/pull/15684): Añade soporte de resolución 1080p para Seedance 2.5
 * [**ByteDance Seedance 2.5 Extend**](https://github.com/Comfy-Org/ComfyUI/pull/15579): Añade el tipo de tarea extend para la continuación de vídeos
 
 </Update>
 
 <Update label="v0.33.1" description="13 de agosto de 2026">
 
 **Nuevo soporte para modelos de código abierto**
 * [**MiniMax Music 3**](https://links.comfy.org/4zgmIJ9): generación nativa de texto a música a partir de descripciones y letras, de hasta 5 minutos
 * [**Anima tunes**](https://github.com/Comfy-Org/ComfyUI/pull/15555): soporte para checkpoints de Anima tunes con bloques adicionales
 
 **Actualizaciones de nodos de socios**
 * [**MiniMax H3 Context IR & Regenerate**](https://github.com/Comfy-Org/ComfyUI/pull/15471): se añadieron el potenciador de prompts Context IR y la función Regenerate a los nodos 2K
 * [**Bria GenFill, Eraser, Expand & Upscale**](https://github.com/Comfy-Org/ComfyUI/pull/15572): se añadieron los nodos de Relleno Generativo, Borrador, Expandir Imagen y Aumentar Resolución
 
 </Update>
 
 <Update label="v0.32.0" description="11 de agosto de 2026">
 
 **Nuevo soporte para modelos de código abierto**
 * [**LTX 2.5**](https://links.comfy.org/cloLTX2_5): Soporte nativo con STG, CFG dual y predicción de duración
 
 **Actualizaciones de nodos asociados**
 * [**Qwen Image 3.0**](https://links.comfy.org/45W0b6L): Añade nodos de texto a imagen y edición para 3.0 y 3.0 Pro
 * [**LTX 2.5**](https://links.comfy.org/3TV1HDB): Nodos asociados de texto, imagen y audio a video
 * [**Grok Imagine Image 2.0**](https://github.com/Comfy-Org/ComfyUI/pull/15496): Añade soporte para el modelo grok-imagine-image-2.0
 
 </Update>
 
 <Update label="v0.31.0" description="7 de agosto de 2026">
 
 **Nuevo soporte para modelos de código abierto**
 * [**Wan-Animate2**](https://links.comfy.org/3TAWMHV): Wan-Animate2 nativo con controles de pose/referencia y caché opcional de rama de pose
 
 **Actualizaciones de nodos partner**
 * [**ByteDance SeeDance 2.5**](https://links.comfy.org/3RPYzbv): Se añadió soporte para el modelo ByteDance SeeDance 2.5
 * [**Seedream 5.0 Layer Separation**](https://links.comfy.org/4hgRB9L): Soporte de separación de capas de ByteDance Seedream 5.0 Pro
 * [**Reve obsoleto**](https://github.com/Comfy-Org/ComfyUI/pull/15331): Los nodos partner de Reve han quedado obsoletos
 
 **Actualizaciones de nuevos nodos**
 * [**Create Layered Image**](https://github.com/Comfy-Org/ComfyUI/pull/15317): Composición de capas estilo PSD con modos de fusión y estado de capa
 * [**Add Layer**](https://github.com/Comfy-Org/ComfyUI/pull/15317): Añadir una capa a una pila de capas del compositor
 * [**Layers From Bounding Boxes**](https://github.com/Comfy-Org/ComfyUI/pull/15317): Construir capas del compositor a partir de entradas de cuadros delimitadores
 * [**WanAnimate2ToVideo**](https://github.com/Comfy-Org/ComfyUI/pull/15362): Acondicionamiento de Wan-Animate2 con controles de pose y referencia
 * [**WanAnimate2Cache**](https://github.com/Comfy-Org/ComfyUI/pull/15362): Caché opcional de rama de pose para acelerar Wan-Animate2
 
 </Update>
 
 <Update label="v0.30.2" description="5 de agosto de 2026">
 
 **Actualizaciones de nodos de socios**
 * [**BFL Flux 3 Video**](https://links.comfy.org/3TVdfGP): Añadido soporte para el modelo de video Flux 3
 * [**TopazAI Bloom 2 y Wonder 3.5**](https://github.com/Comfy-Org/ComfyUI/pull/15294): Añadido soporte para los modelos Bloom 2 y Wonder 3.5
 * [**Kling EOL**](https://github.com/Comfy-Org/ComfyUI/pull/15249): Eliminados los modelos Kling heredados retirados y la API de Virtual Try-On
 
 </Update>
 
 <Update label="v0.30.1" description="3 de agosto de 2026">
 
 **Rendimiento y estabilidad**
 * [**Frontend**](https://github.com/Comfy-Org/ComfyUI/pull/15244): Se actualizó comfyui-frontend-package a 1.47.12
 
 </Update>
 
 <Update label="v0.30.0" description="3 de agosto de 2026">
 
 **Nuevo soporte de modelos de código abierto**
 * [**MiniMax-H3**](https://links.comfy.org/3TuT9TO): Añade soporte para el modelo de audio y vídeo MiniMax-H3
 * [**Pruna LTX VAE**](https://github.com/Comfy-Org/ComfyUI/pull/15129): Soporte de PrunaVAED para una decodificación LTX 2.3 más rápida
 
 **Nuevas actualizaciones de nodos**
 * [**Save Video CRF**](https://github.com/Comfy-Org/ComfyUI/pull/15191): Añade la opción CRF al nodo Save Video
 * [**SaveText CSV**](https://github.com/Comfy-Org/ComfyUI/pull/15217): El nodo SaveText ahora puede guardar la salida en .csv
 * [**VAEDecodeAudio**](https://github.com/Comfy-Org/ComfyUI/pull/15211): Gestiona la decodificación de latentes de audio anidados
 
 **Actualizaciones de nodos de socios**
 * [**xAI Grok Imagine Video 1.5**](https://github.com/Comfy-Org/ComfyUI/pull/15197): Nodos xAI actualizados para dar soporte a grok-imagine-video-1.5
 * [**MiniMax H3 768P**](https://github.com/Comfy-Org/ComfyUI/pull/15227): Añade la resolución 768P para los nodos MiniMax H3
 
 </Update>
 
 <Update label="v0.29.2" description="31 de julio de 2026">
 
 **Actualizaciones de nodos de socios**
 * [**Ideogram P-Image**](https://github.com/Comfy-Org/ComfyUI/pull/15154): Añade soporte para el modelo Ideogram P-Image
 * [**MiniMax H3**](https://github.com/Comfy-Org/ComfyUI/pull/15167): Añade soporte para el modelo MiniMax H3
 
 </Update>
 
 <Update label="v0.29.0" description="29 de julio de 2026">
 
 **Nuevo soporte para modelos de código abierto**
 * [**JoyImageEdit**](https://github.com/Comfy-Org/ComfyUI/pull/14428): Soporte nativo para el modelo JoyImageEdit
 * [**Gemma4 12B**](https://github.com/Comfy-Org/ComfyUI/pull/14304): Soporte para el modelo Gemma4 12B
 * [**MageFlow**](https://github.com/Comfy-Org/ComfyUI/pull/15026): Soporte para el modelo MageFlow
 * [**Uni3C ControlNet**](https://github.com/Comfy-Org/ComfyUI/pull/14946): Soporte nativo para Uni3C ControlNet en modelos Wan
 * [**Anima**](https://github.com/Comfy-Org/ComfyUI/pull/14954): Soporte para el modelo de control Anima llite
 
 **Nuevas actualizaciones de nodos**
 * [**Procesamiento de video para Trainer**](https://github.com/Comfy-Org/ComfyUI/pull/13588): Nodos de procesamiento de video y soporte de video en el nodo de procesamiento de imágenes
 
 **Actualizaciones de nodos de socios**
 * [**Recraft V4.1**](https://github.com/Comfy-Org/ComfyUI/pull/15105): Modelos Recraft V4.1 y nodos de edición de imágenes V4
 * [**ByteDance audio multilingüe**](https://github.com/Comfy-Org/ComfyUI/pull/15034): Soporte para el modelo ByteDance audio-1.0-multilingual
 * [**OpenRouter**](https://github.com/Comfy-Org/ComfyUI/pull/15021): Nuevos modelos de OpenRouter, incluido Claude Opus 5
 * [**Anthropic**](https://github.com/Comfy-Org/ComfyUI/pull/15023): Nuevos modelos de Anthropic, incluido Claude Opus 5
 
 </Update>
 
 <Update label="v0.28.3" description="22 de julio de 2026">
 
 **Actualizaciones de nodos partner**
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/15014): Pasa videos como datos en línea en el nodo partner de Gemini Omni
 * [**OpenRouter**](https://github.com/Comfy-Org/ComfyUI/pull/15021): Nuevos modelos de OpenRouter
 * [**Anthropic**](https://github.com/Comfy-Org/ComfyUI/pull/15023): Nuevos modelos de Anthropic
 
 </Update>
 
 <Update label="v0.28.2" description="July 18, 2026">
 
 **Actualizaciones de nodos asociados**
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/14984): Cambió el modelo Omni a la API de Interactions
 
 </Update>
 
 <Update label="v0.28.1" description="17 de julio de 2026">
 
 **Actualizaciones de nodos asociados**
 * [**GPT-5.6**](https://github.com/Comfy-Org/ComfyUI/pull/14957): Se añadieron los modelos GPT-5.6 Sol, Terra y Luna
 * [**Gemini 3.5 Flash**](https://github.com/Comfy-Org/ComfyUI/pull/14972): Se añadió Gemini 3.5 Flash al nodo Gemini LLM
 * [**HeyGen**](https://github.com/Comfy-Org/ComfyUI/pull/14958): Nodos de Video de Avatar, Foto Parlante, Crear Avatar, Traducir Video y Texto a Voz
 
 </Update>
 
 <Update label="v0.28.0" description="July 15, 2026">
 
 **Nuevo soporte para modelos de código abierto**
 * [**SeedVR2**](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/utility_seedvr2_3b_int8_upscale_image.json): Escalado nativo de imágenes y vídeos con SeedVR2
 * [**PID 1.5**](https://github.com/Comfy-Org/ComfyUI/pull/14894): Se añadió soporte para el modelo PixelDiT PID 1.5
 * [**int4 Models**](https://github.com/Comfy-Org/ComfyUI/pull/14859): Soporte para convrot int4 y optimizaciones Turing int8/int4
 
 **Nuevas actualizaciones de nodos**
 * [**Save 3D (Advanced)**](https://github.com/Comfy-Org/ComfyUI/pull/14701): Nodos Guardar 3D (Avanzado), Guardar Splat y Guardar Nube de Puntos
 * [**SaveText**](https://github.com/Comfy-Org/ComfyUI/pull/14102): Guardar salidas de texto en .txt, .md o .json
 * [**Text Overlay**](https://github.com/Comfy-Org/ComfyUI/pull/14610): Superposición de texto para imágenes y vídeo
 * [**Text**](https://github.com/Comfy-Org/ComfyUI/pull/14870): Nodo Text integrado rehabilitado
 
 **Actualizaciones de nodos asociados**
 * [**sync.so sync-3**](https://github.com/Comfy-Org/ComfyUI/pull/14928): Soporte para sincronización de labios y talking-image con sync.so sync-3
 * [**Seedream thinking control**](https://github.com/Comfy-Org/ComfyUI/pull/14853): Opción para desactivar el pensamiento en nodos Seedream
 * [**Gemini Image**](https://github.com/Comfy-Org/ComfyUI/pull/14917): Modelos de previsualización de Gemini Image trasladados a versiones de lanzamiento
 
 </Update>
 
 <Update label="v0.27.1" description="July 8, 2026">
 
 **Actualizaciones de nodos asociados**
 * [**Seedream 5.0 Pro**](https://links.comfy.org/4ydIJYi): Soporte para modelo de generación de imágenes de ByteDance añadido.
 * [**Seed Audio 1.0**](https://github.com/Comfy-Org/ComfyUI/pull/14731): Voz, música, efectos de sonido, diálogo multi-orador.
 * [**Ideogram V1/V2 obsoleto**](https://github.com/Comfy-Org/ComfyUI/pull/14712): Nodos asociados heredados de Ideogram obsoletos.
 * [**StabilityAI eliminado temporalmente**](https://github.com/Comfy-Org/ComfyUI/pull/14737): Nodos asociados de StabilityAI eliminados temporalmente.
 
 </Update>
 
 <Update label="v0.27.0" description="30 de junio de 2026">
 
 **Soporte Int8**
 * [**modelos convrot int8**](https://github.com/Comfy-Org/ComfyUI/pull/14636): Soporte para modelos convrot cuantificados int8 con inferencia más rápida en GPUs Turing
 
 **Actualizaciones de Nodos Asociados**
 * [**Nano Banana 2 Lite**](https://github.com/Comfy-Org/ComfyUI/pull/14693): El modelo ligero de generación de imágenes de Google
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/14695): Nodo de generación de video impulsado por Gemini
 
 **Actualizaciones de Nuevos Nodos**
 * [**Seed Node**](https://github.com/Comfy-Org/ComfyUI/pull/14616): Simplifica la gestión de semillas para flujos de trabajo reproducibles
 * [**Advanced Krea 2 Model Merging**](https://github.com/Comfy-Org/ComfyUI/pull/14621): Fusiona modelos Krea 2
 * [**Bounding Box Canvas**](https://github.com/Comfy-Org/ComfyUI/pull/14537): Lienzo con soporte de prompt JSON de Ideogram
 * [**ConditioningMultiply**](https://github.com/Comfy-Org/ComfyUI/pull/14686): Multiplica valores de condicionamiento
 * [**GLSL Update**](https://github.com/Comfy-Org/ComfyUI/pull/13195): Nodo GLSL actualizado para usar la biblioteca ANGLE
 
 </Update>
 
 <Update label="v0.26.2" description="25 de junio de 2026">
 
 **Actualizaciones de nodos Partner**
 * [**SeeDance 2.0 Mini**](https://links.comfy.org/4eOUnjJ): Se añadió ByteDance SeeDance 2.0 Mini en los nodos Text-to-Video, First-Last-Frame y Reference
 
 </Update>
 
 <Update label="v0.26.1" description="24 de junio de 2026">
 
 **Actualizaciones de nodos asociados**
 * [**SeeDance 2.0**](https://github.com/Comfy-Org/ComfyUI/pull/14614): Se añadió compatibilidad con resolución 4K para ByteDance SeeDance 2.0
 * [**HappyHorse 1.1**](https://links.comfy.org/3SoBFYI): Se añadió compatibilidad con el modelo HappyHorse 1.1 de Alibaba
 * [**Grok Video**](https://github.com/Comfy-Org/ComfyUI/pull/14612): Se añadió compatibilidad con 1080p para la generación de video Grok
 
 </Update>
 
 <Update label="v0.26.0" description="23 de junio de 2026">
 
 **Nuevo soporte para modelos de código abierto**
 * [**Krea2**](https://links.comfy.org/4eC8714): Modelo de imagen de código abierto de Krea con amplio rango estético y estilístico
 * [**Boogu-Image**](https://github.com/Comfy-Org/ComfyUI/pull/14523): Texto a imagen con codificador Qwen3-VL-8B y edición de imágenes
 * [**Generación de texto Qwen3-VL**](https://github.com/Comfy-Org/ComfyUI/pull/14298): Generación de texto consciente de la imagen con modelos 4B y 8B
 * [**SCAIL-2 Multireferencia**](https://github.com/Comfy-Org/ComfyUI/pull/14509): Reemplazo de personajes con múltiples referencias SCAIL-2
 
 **Nuevas actualizaciones de nodos**
 * [**Ventanas de contexto LTX2**](https://github.com/Comfy-Org/ComfyUI/pull/13325): Muestreo de ventana de contexto para video+audio LTX-2 con empalme IC-LoRA
 * [**Socket de salida de nodos de guardado**](https://github.com/Comfy-Org/ComfyUI/pull/13866): Socket de salida en nodos de guardado para cableado descendente
 
 **Actualizaciones de nodos de socios**
 * [**Luma Rays 3.2**](https://github.com/Comfy-Org/ComfyUI/pull/14540): Siete nodos para la generación de video Luma Ray 3.2
 
 </Update>
 
 <Update label="v0.25.1" description="16 de junio de 2026">
 
 **Actualizaciones de Nodos Asociados**
 * [**Kling V3-Turbo**](https://github.com/Comfy-Org/ComfyUI/pull/14528): Modelo de generación de texto a video
 
 </Update>
 
 <Update label="v0.25.0" description="June 16, 2026">
 
 **Soporte para nuevos modelos de código abierto**
 * [**Depth Anything 3**](https://github.com/Comfy-Org/ComfyUI/pull/13853): Modelo de estimación de profundidad monocular
 * [**SCAIL-2**](https://github.com/Comfy-Org/ComfyUI/pull/14373): Modelo de reemplazo de personajes para edición de video
 * [**Bernini-R**](https://github.com/Comfy-Org/ComfyUI/pull/14216): Soporte para modelo de video Wan
 
 **Actualizaciones de nodos de socios**
 * [**Gemini Text Node**](https://github.com/Comfy-Org/ComfyUI/pull/14299): Nodo de socio LLM
 * [**Runway Aleph2**](https://github.com/Comfy-Org/ComfyUI/pull/14306): Nodo de generación de video
 
 </Update>
 
 <Update label="v0.24.1" description="3 de junio de 2026">
 
 **Actualizaciones de nodos asociados**
 * **Bria Green Background**: Nodo de eliminación de fondo verde de imagen.
 * **Bria Transparent Video Background**: Eliminación de fondo de video con salida de canal alfa.
 * **Krea 2 Medium Turbo**: Modelo más rápido de generación de imágenes de tamaño mediano.
 * **Entrada de semilla al nodo Flux Erase**: Añade parámetro de semilla para resultados de borrado repetibles.
 
 </Update>
 
 <Update label="v0.23.0" description="1 de junio de 2026">
 
 Nuevo soporte para modelos de código abierto
 * [**Microsoft Lens**](https://github.com/Comfy-Org/ComfyUI/pull/14077): Modelo de texto a imagen con codificador GPT-OSS-20B
 * [**NVIDIA PixelDiT & PiD**](https://github.com/Comfy-Org/ComfyUI/pull/14103): Modelos de texto a imagen que incluyen PixelDiT 1300M y la familia PiD
 * [**TripoSplat**](https://github.com/Comfy-Org/ComfyUI/pull/14210): Modelo de splats gaussianos de imagen a 3D
 * [**MediaPipe Face Detection**](https://github.com/Comfy-Org/ComfyUI/pull/14009): Detección de rostros y puntos de referencia sin dependencias
 
 Actualizaciones de nodos asociados
 * **OpenRouter LLM Node**: Nodo asociado para la API de modelo de lenguaje de OpenRouter
 
 </Update>
 
 <Update label="v0.22.3" description="May 20, 2026">
 
 **Actualizaciones de nodos asociados**
 * **Nodos de imagen Krea2**: Se añadieron nuevos nodos de generación de imágenes para flujos de trabajo de Krea2
 * **SeeDance 2**: Se mejoró la carga de referencias de video mediante memoryview
 
 </Update>
 
 <Update label="v0.22.2" description="20 de mayo de 2026">
 
 **Actualizaciones de nodos asociados**
 * **Rodin2.5**: Nodo de generación de malla de calidad
 * **Rodin2.5**: Corrección en el manejo del parámetro semilla
 
 </Update>
 
 <Update label="v0.22.1" description="20 de mayo de 2026">
 
 * **OpenRouter LLM node**: Nuevo nodo asociado para el acceso a la API de OpenRouter.
 * **Anthropic node**: Añadido widget de razonamiento para mostrar el pensamiento paso a paso.
 * **ByteDance2Reference node**: Añadido widget de escalado automático para las salidas de video.
 * **Grok LLM**: Corregido un problema al pasar imágenes al modelo.
 
 </Update>

.github/scripts/cms/staging/cloud/fr/changelog/index.mdx

@@ -4,255 +4,262 @@
 ---
 
 
+<Update label="v0.34.3" description="2 septembre 2026">
+
+**Mises à jour des nœuds partenaires**
+* [**MiniMax H3 Max**](https://cloud.comfy.org/?template=api_minimax_h3_max_r2v) : Ajout de la prise en charge du modèle H3 Max
+
+</Update>
+
 <Update label="v0.34.2" description="27 août 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Gemini Omni 1.1 Flash**](https://links.comfy.org/4qIN5DH) : Génération vidéo plus rapide et édition conversationnelle, avec sortie 4K et extension vidéo
 
 </Update>
 
 <Update label="v0.34.0" description="26 août 2026">
 
 **Nouveau support de modèles open-source**
 * [**Pixal3D & TRELLIS2**](https://cloud.comfy.org/?template=3d_pixal3d_trellis2_image_to_model) : Génération de maillages 3D et de textures avec remaillage, décimation, dépliage UV, cuisson de textures
 * [**SAM 3D Body**](https://cloud.comfy.org/?template=utility_sam3d_body) : Détection et segmentation du corps 3D sans dépendance, avec rendu de pose et export d'animation
 * [**TAESD H3**](https://github.com/Comfy-Org/ComfyUI/pull/15695) : Décodeur VAE léger pour des aperçus plus rapides
 
 **Mises à jour des nœuds partenaires**
 * [**PixVerse V6**](https://github.com/Comfy-Org/ComfyUI/pull/15880) : Ajout des nœuds V6 texte-vers-vidéo, image-vers-vidéo, première-dernière frame, extension, fusion avec audio natif
 * [**Tripo**](https://github.com/Comfy-Org/ComfyUI/pull/15851) : Suppression du nœud de modèle Refine Draft retiré et de la version de modèle v2.0
 
 </Update>
 
 <Update label="v0.33.4" description="August 24, 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Wan 3.0**](https://cloud.comfy.org/?template=api_wan3_0_i2v) : Génération de texte, d'image et de référence vers vidéo avec audio natif jusqu'à 30 secondes
 * [**Meshy-7**](https://cloud.comfy.org/?template=api_meshy7_image_to_model) : Ajoute meshy-7 et le mode ultra aux nœuds texte, image et multi-image vers 3D
 * [**ByteDance vCube Video Enhance**](https://cloud.comfy.org/?template=api_bytedance_vcube_video_enhance) : Améliore et restaure les vidéos jusqu'en 8K avec interpolation d'images optionnelle
 * [**Kling v2**](https://github.com/Comfy-Org/ComfyUI/pull/15676) : Supprime kling-v2 de Kling Image Generation avant son retrait le 15 septembre
 
 </Update>
 
 <Update label="v0.33.3" description="20 août 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Flux Video Upscale**](https://github.com/Comfy-Org/ComfyUI/pull/15753) : Agrandissez la vidéo de 1,5x à 3x avec la super-résolution FLUX
 * [**ByteDance Seedream Fast Mode**](https://github.com/Comfy-Org/ComfyUI/pull/15750) : Mode rapide d'optimisation de prompt pour l'image-à-image Seedream 5.0 Pro
 * [**Gemini 3.7 Flash**](https://github.com/Comfy-Org/ComfyUI/pull/15688) : Gemini 3.7 Flash ajouté au nœud de texte Gemini
 * [**Ideogram & Pruna P-Image**](https://github.com/Comfy-Org/ComfyUI/pull/15751) : Ideogram P-Image renommé pour créditer Pruna
 
 </Update>
 
 <Update label="v0.33.2" description="17 août 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Fish Audio**](https://github.com/Comfy-Org/ComfyUI/pull/15612) : Ajoute des nœuds de synthèse vocale, de reconnaissance vocale, de sélection de voix et de clonage instantané de voix
 * [**ByteDance Seedance 2.5 1080p**](https://github.com/Comfy-Org/ComfyUI/pull/15684) : Ajoute la prise en charge de la résolution 1080p pour Seedance 2.5
 * [**ByteDance Seedance 2.5 Extend**](https://github.com/Comfy-Org/ComfyUI/pull/15579) : Ajoute le type de tâche « extend » pour la continuation vidéo
 
 </Update>
 
 <Update label="v0.33.1" description="13 août 2026">
 
 **Nouveau support de modèles open-source**
 * [**MiniMax Music 3**](https://links.comfy.org/4zgmIJ9) : Texte-vers-musique natif à partir d'une légende et de paroles, jusqu'à 5 minutes
 * [**Anima tunes**](https://github.com/Comfy-Org/ComfyUI/pull/15555) : Prise en charge des checkpoints Anima tunes avec blocs supplémentaires
 
 **Mises à jour des nœuds partenaires**
 * [**MiniMax H3 Context IR & Regenerate**](https://github.com/Comfy-Org/ComfyUI/pull/15471) : Ajout de l'améliorateur de prompt Context IR et de Regenerate aux nœuds 2K
 * [**Bria GenFill, Eraser, Expand & Upscale**](https://github.com/Comfy-Org/ComfyUI/pull/15572) : Ajout des nœuds Generative Fill, Eraser, Expand Image et Increase Resolution
 
 </Update>
 
 <Update label="v0.32.0" description="11 août 2026">
 
 **Nouveau support de modèles open-source**
 * [**LTX 2.5**](https://links.comfy.org/cloLTX2_5): Prise en charge native avec STG, double CFG, prédiction de durée
 
 **Mises à jour des nœuds partenaires**
 * [**Qwen Image 3.0**](https://links.comfy.org/45W0b6L): Ajoute les nœuds 3.0 et 3.0 Pro de texte-vers-image et d'édition
 * [**LTX 2.5**](https://links.comfy.org/3TV1HDB): Nœuds partenaires texte, image et audio vers vidéo
 * [**Grok Imagine Image 2.0**](https://github.com/Comfy-Org/ComfyUI/pull/15496): Ajoute la prise en charge du modèle grok-imagine-image-2.0
 
 </Update>
 
 <Update label="v0.31.0" description="7 août 2026">
 
 **Nouveau support de modèles open source**
 * [**Wan-Animate2**](https://links.comfy.org/3TAWMHV) : Wan-Animate2 natif avec contrôles de pose/référence et cache de branche de pose optionnel
 
 **Mises à jour des nœuds partenaires**
 * [**ByteDance SeeDance 2.5**](https://links.comfy.org/3RPYzbv) : Ajout du support du modèle ByteDance SeeDance 2.5
 * [**Seedream 5.0 Layer Separation**](https://links.comfy.org/4hgRB9L) : Support de la séparation des calques ByteDance Seedream 5.0 Pro
 * [**Reve déprécié**](https://github.com/Comfy-Org/ComfyUI/pull/15331) : Les nœuds partenaires Reve ont été dépréciés
 
 **Nouvelles mises à jour de nœuds**
 * [**Create Layered Image**](https://github.com/Comfy-Org/ComfyUI/pull/15317) : Composition de calques de style PSD avec modes de fusion et état de calque
 * [**Add Layer**](https://github.com/Comfy-Org/ComfyUI/pull/15317) : Ajouter un calque dans une pile de calques de composition
 * [**Layers From Bounding Boxes**](https://github.com/Comfy-Org/ComfyUI/pull/15317) : Construire des calques de composition à partir d'entrées de boîtes englobantes
 * [**WanAnimate2ToVideo**](https://github.com/Comfy-Org/ComfyUI/pull/15362) : Conditionnement Wan-Animate2 avec contrôles de pose et de référence
 * [**WanAnimate2Cache**](https://github.com/Comfy-Org/ComfyUI/pull/15362) : Cache de branche de pose optionnel pour accélérer Wan-Animate2
 
 </Update>
 
 <Update label="v0.30.2" description="5 août 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**BFL Flux 3 Video**](https://links.comfy.org/3TVdfGP) : Ajout de la prise en charge du modèle vidéo Flux 3
 * [**TopazAI Bloom 2 & Wonder 3.5**](https://github.com/Comfy-Org/ComfyUI/pull/15294) : Ajout de la prise en charge des modèles Bloom 2 et Wonder 3.5
 * [**Kling EOL**](https://github.com/Comfy-Org/ComfyUI/pull/15249) : Suppression des anciens modèles Kling et de l'API Virtual Try-On
 
 </Update>
 
 <Update label="v0.30.1" description="3 août 2026">
 
 **Performance et stabilité**
 * [**Frontend**](https://github.com/Comfy-Org/ComfyUI/pull/15244): Mise à jour de comfyui-frontend-package vers 1.47.12
 
 </Update>
 
 <Update label="v0.30.0" description="3 août 2026">
 
 **Prise en charge de nouveaux modèles open-source**
 * [**MiniMax-H3**](https://links.comfy.org/3TuT9TO): Ajoute la prise en charge du modèle audio-vidéo pour MiniMax-H3
 * [**Pruna LTX VAE**](https://github.com/Comfy-Org/ComfyUI/pull/15129): Prise en charge de PrunaVAED pour un décodage LTX 2.3 plus rapide
 
 **Nouvelles mises à jour de nœuds**
 * [**Save Video CRF**](https://github.com/Comfy-Org/ComfyUI/pull/15191): Ajoute l'option CRF au nœud Save Video
 * [**SaveText CSV**](https://github.com/Comfy-Org/ComfyUI/pull/15217): Le nœud SaveText peut désormais enregistrer la sortie .csv
 * [**VAEDecodeAudio**](https://github.com/Comfy-Org/ComfyUI/pull/15211): Gère le décodage des latents audio imbriqués
 
 **Mises à jour des nœuds partenaires**
 * [**xAI Grok Imagine Video 1.5**](https://github.com/Comfy-Org/ComfyUI/pull/15197): Nœuds xAI mis à jour pour la prise en charge de grok-imagine-video-1.5
 * [**MiniMax H3 768P**](https://github.com/Comfy-Org/ComfyUI/pull/15227): Ajoute la résolution 768P pour les nœuds MiniMax H3
 
 </Update>
 
 <Update label="v0.29.2" description="31 juillet 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Ideogram P-Image**](https://github.com/Comfy-Org/ComfyUI/pull/15154) : Ajoute la prise en charge du modèle Ideogram P-Image
 * [**MiniMax H3**](https://github.com/Comfy-Org/ComfyUI/pull/15167) : Ajoute la prise en charge du modèle MiniMax H3
 
 </Update>
 
 <Update label="v0.29.0" description="29 juillet 2026">
 
 **Nouveau support de modèles open-source**
 * [**JoyImageEdit**](https://github.com/Comfy-Org/ComfyUI/pull/14428) : support natif du modèle JoyImageEdit
 * [**Gemma4 12B**](https://github.com/Comfy-Org/ComfyUI/pull/14304) : support du modèle Gemma4 12B
 * [**MageFlow**](https://github.com/Comfy-Org/ComfyUI/pull/15026) : support du modèle MageFlow
 * [**Uni3C ControlNet**](https://github.com/Comfy-Org/ComfyUI/pull/14946) : support natif d'Uni3C ControlNet pour les modèles Wan
 * [**Anima**](https://github.com/Comfy-Org/ComfyUI/pull/14954) : support du modèle de contrôle Anima llite
 
 **Nouvelles mises à jour de nœuds**
 * [**Traitement vidéo Trainer**](https://github.com/Comfy-Org/ComfyUI/pull/13588) : nœuds de traitement vidéo et prise en charge vidéo du nœud de traitement d'images
 
 **Mise à jour des nœuds partenaires**
 * [**Recraft V4.1**](https://github.com/Comfy-Org/ComfyUI/pull/15105) : modèles Recraft V4.1 et nœuds d'édition d'images V4
 * [**ByteDance audio multilingue**](https://github.com/Comfy-Org/ComfyUI/pull/15034) : support du modèle ByteDance audio-1.0-multilingual
 * [**OpenRouter**](https://github.com/Comfy-Org/ComfyUI/pull/15021) : nouveaux modèles OpenRouter, y compris Claude Opus 5
 * [**Anthropic**](https://github.com/Comfy-Org/ComfyUI/pull/15023) : nouveaux modèles Anthropic, y compris Claude Opus 5
 
 </Update>
 
 <Update label="v0.28.3" description="July 22, 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/15014): Transmettre des vidéos en tant que données intégrées dans le nœud partenaire Gemini Omni
 * [**OpenRouter**](https://github.com/Comfy-Org/ComfyUI/pull/15021): Nouveaux modèles OpenRouter
 * [**Anthropic**](https://github.com/Comfy-Org/ComfyUI/pull/15023): Nouveaux modèles Anthropic
 
 </Update>
 
 <Update label="v0.28.2" description="18 juillet 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/14984) : Passage du modèle Omni à l'API Interactions
 
 </Update>
 
 <Update label="v0.28.1" description="17 juillet 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**GPT-5.6**](https://github.com/Comfy-Org/ComfyUI/pull/14957) : Ajout des modèles GPT-5.6 Sol, Terra et Luna
 * [**Gemini 3.5 Flash**](https://github.com/Comfy-Org/ComfyUI/pull/14972) : Ajout de Gemini 3.5 Flash au nœud LLM Gemini
 * [**HeyGen**](https://github.com/Comfy-Org/ComfyUI/pull/14958) : Nœuds Avatar Video, Talking Photo, Create Avatar, Video Translate, Text to Speech
 
 </Update>
 
 <Update label="v0.28.0" description="July 15, 2026">
 
 **Nouveau support de modèles open-source**
 * [**SeedVR2**](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/utility_seedvr2_3b_int8_upscale_image.json) : Mise à l'échelle native d'images et de vidéos SeedVR2
 * [**PID 1.5**](https://github.com/Comfy-Org/ComfyUI/pull/14894) : Ajout du support du modèle PixelDiT PID 1.5
 * [**int4 Models**](https://github.com/Comfy-Org/ComfyUI/pull/14859) : Support convrot int4 et optimisations Turing int8/int4
 
 **Nouvelles mises à jour de nœuds**
 * [**Save 3D (Advanced)**](https://github.com/Comfy-Org/ComfyUI/pull/14701) : Nœuds Save 3D (Advanced), Save Splat et Save Point Cloud
 * [**SaveText**](https://github.com/Comfy-Org/ComfyUI/pull/14102) : Enregistrement des sorties texte en .txt, .md ou .json
 * [**Text Overlay**](https://github.com/Comfy-Org/ComfyUI/pull/14610) : Superposition de texte pour images et vidéos
 * [**Text**](https://github.com/Comfy-Org/ComfyUI/pull/14870) : Nœud Text intégré réactivé
 
 **Mises à jour des nœuds partenaires**
 * [**sync.so sync-3**](https://github.com/Comfy-Org/ComfyUI/pull/14928) : Support sync.so sync-3 lip-sync et talking-image
 * [**Seedream thinking control**](https://github.com/Comfy-Org/ComfyUI/pull/14853) : Option pour désactiver le raisonnement sur les nœuds Seedream
 * [**Gemini Image**](https://github.com/Comfy-Org/ComfyUI/pull/14917) : Les modèles d'aperçu Gemini Image sont passés aux versions de publication
 
 </Update>
 
 <Update label="v0.27.1" description="8 juillet 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**Seedream 5.0 Pro**](https://links.comfy.org/4ydIJYi): Ajout de la prise en charge du modèle de génération d'images ByteDance.
 * [**Seed Audio 1.0**](https://github.com/Comfy-Org/ComfyUI/pull/14731): Parole, musique, effets sonores, dialogue multi-locuteur.
 * [**Ideogram V1/V2 obsolète**](https://github.com/Comfy-Org/ComfyUI/pull/14712): Les nœuds partenaires Ideogram hérités sont obsolètes.
 * [**StabilityAI temporairement retiré**](https://github.com/Comfy-Org/ComfyUI/pull/14737): Les nœuds partenaires StabilityAI ont été temporairement retirés.
 
 </Update>
 
 <Update label="v0.27.0" description="30 juin 2026">
 
 **Support Int8**
 * [**modèles convrot int8**](https://github.com/Comfy-Org/ComfyUI/pull/14636) : Prise en charge des modèles convrot quantifiés en int8 avec inférence plus rapide sur les GPU Turing
 
 **Mises à jour des nœuds partenaires**
 * [**Nano Banana 2 Lite**](https://github.com/Comfy-Org/ComfyUI/pull/14693) : Modèle léger de génération d'images de Google
 * [**Gemini Video Omni**](https://github.com/Comfy-Org/ComfyUI/pull/14695) : Nœud de génération vidéo propulsé par Gemini
 
 **Mises à jour des nouveaux nœuds**
 * [**Nœud Seed**](https://github.com/Comfy-Org/ComfyUI/pull/14616) : Simplifie la gestion des graines pour des workflows reproductibles
 * [**Fusion avancée de modèles Krea 2**](https://github.com/Comfy-Org/ComfyUI/pull/14621) : Fusionne les modèles Krea 2
 * [**Canvas Bounding Box**](https://github.com/Comfy-Org/ComfyUI/pull/14537) : Canvas avec prise en charge des invites JSON Ideogram
 * [**ConditioningMultiply**](https://github.com/Comfy-Org/ComfyUI/pull/14686) : Multiplie les valeurs de conditionnement
 * [**Mise à jour GLSL**](https://github.com/Comfy-Org/ComfyUI/pull/13195) : Nœud GLSL mis à jour pour utiliser la bibliothèque ANGLE
 
 </Update>
 
 <Update label="v0.26.2" description="25 juin 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**SeeDance 2.0 Mini**](https://links.comfy.org/4eOUnjJ) : Ajout de ByteDance SeeDance 2.0 Mini sur les nœuds Text-to-Video, First-Last-Frame et Reference
 
 </Update>
 
 <Update label="v0.26.1" description="24 juin 2026">
 
 **Mises à jour des nœuds partenaires**
 * [**SeeDance 2.0**](https://github.com/Comfy-Org/ComfyUI/pull/14614): Ajout du support de la résolution 4K pour ByteDance SeeDance 2.0
 * [**HappyHorse 1.1**](https://links.comfy.org/3SoBFYI): Ajout du support du modèle HappyHorse 1.1 d'Alibaba
 * [**Grok Video**](https://github.com/Comfy-Org/ComfyUI/pull/14612): Ajout du support 1080p pour la génération vidéo Grok
 
 </Update>
 
 <Update label="v0.26.0" description="June 23, 2026">
 
 **Prise en charge de nouveaux modèles open source**
 * [**Krea2**](https://links.comfy.org/4eC8714) : Modèle d'image open source de Krea à large gamme esthétique et stylistique
 * [**Boogu-Image**](https://github.com/Comfy-Org/ComfyUI/pull/14523) : Génération texte-vers-image avec l'encodeur Qwen3-VL-8B et édition d'image
 * [**Qwen3-VL Text Generation**](https://github.com/Comfy-Org/ComfyUI/pull/14298) : Génération de texte sensible à l'image avec les modèles 4B et 8B
 * [**SCAIL-2 Multireference**](https://github.com/Comfy-Org/ComfyUI/pull/14509) : Remplacement de personnage SCAIL-2 multi-références
 
 **Nouvelles mises à jour de nœuds**
 * [**LTX2 Context Windows**](https://github.com/Comfy-Org/ComfyUI/pull/13325) : Échantillonnage par fenêtre de contexte pour vidéo+audio LTX-2 avec épissage IC-LoRA

.github/scripts/cms/staging/cloud/ja/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2026年9月2日">
+
+**パートナーノードの更新**
+* [**MiniMax H3 Max**](https://cloud.comfy.org/?template=api_minimax_h3_max_r2v)：H3 Max モデルのサポートが追加されました
+
+</Update>
+
 <Update label="v0.34.2" description="2026年8月27日">
 
 **パートナーノードの更新**

.github/scripts/cms/staging/cloud/ko/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2026년 9월 2일">
+
+**파트너 노드 업데이트**
+* [**MiniMax H3 Max**](https://cloud.comfy.org/?template=api_minimax_h3_max_r2v): H3 Max 모델 지원 추가
+
+</Update>
+
 <Update label="v0.34.2" description="2026년 8월 27일">
 
 **파트너 노드 업데이트**

.github/scripts/cms/staging/cloud/ru/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2 сентября 2026">
+
+**Обновления партнёрских узлов**
+* [**MiniMax H3 Max**](https://cloud.comfy.org/?template=api_minimax_h3_max_r2v): Добавлена поддержка модели H3 Max
+
+</Update>
+
 <Update label="v0.34.2" description="27 августа 2026 г.">
 
 **Обновления партнёрских нод**

.github/scripts/cms/staging/cloud/zh/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2026年9月2日">
+
+**合作伙伴节点更新**
+* [**MiniMax H3 Max**](https://cloud.comfy.org/?template=api_minimax_h3_max_r2v)：添加了 H3 Max 模型支持
+
+</Update>
+
 <Update label="v0.34.2" description="2026年8月27日">
 
 **合作伙伴节点更新**

.github/scripts/cms/staging/en/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="September 2, 2026">
+
+**Partner Node Updates**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025): Adds H3 Max option to text-to-video, first-last-frame, and reference nodes.
+
+</Update>
+
 <Update label="v0.34.2" description="August 27, 2026">
 
 **Partner Node Updates**

.github/scripts/cms/staging/es/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2 de septiembre de 2026">
+
+**Actualizaciones de nodos de socios**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025): Añade la opción H3 Max a los nodos de texto a video, de primer y último fotograma y de referencia.
+
+</Update>
+
 <Update label="v0.34.2" description="27 de agosto de 2026">
 
 **Actualizaciones de nodos de socios**

.github/scripts/cms/staging/fr/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2 septembre 2026">
+
+**Mises à jour des nœuds partenaires**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025) : ajoute l'option H3 Max aux nœuds texte-vers-vidéo, première/dernière image et référence.
+
+</Update>
+
 <Update label="v0.34.2" description="27 août 2026">
 
 **Mises à jour des nœuds partenaires**

.github/scripts/cms/staging/ja/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2026年9月2日">
+
+**パートナーノードの更新**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025): テキストから動画、最初と最後のフレーム、および参照の各ノードに H3 Max オプションを追加します。
+
+</Update>
+
 <Update label="v0.34.2" description="2026年8月27日">
 
 **パートナーノードの更新**

.github/scripts/cms/staging/ko/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2026년 9월 2일">
+
+**파트너 노드 업데이트**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025): 텍스트 기반 비디오 생성, 첫 번째/마지막 프레임, 참조 노드에 H3 Max 옵션을 추가합니다.
+
+</Update>
+
 <Update label="v0.34.2" description="2026년 8월 27일">
 
 **파트너 노드 업데이트**

.github/scripts/cms/staging/ru/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2 сентября 2026">
+
+**Обновления партнёрских узлов**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025): Добавлена опция H3 Max для узлов text-to-video, first-last-frame и reference.
+
+</Update>
+
 <Update label="v0.34.2" description="27 августа 2026 г.">
 
 **Обновления партнёрских нод**

.github/scripts/cms/staging/zh/changelog/index.mdx

@@ -4,6 +4,13 @@ cmsStaging: true
 ---
 
 
+<Update label="v0.34.3" description="2026年9月2日">
+
+**合作伙伴节点更新**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025)：为文生视频、首尾帧和参考节点添加 H3 Max 选项。
+
+</Update>
+
 <Update label="v0.34.2" description="2026年8月27日">
 
 **合作伙伴节点更新**

changelog/index.mdx

@@ -4,6 +4,19 @@ description: "ComfyUI changelog: the latest features, improvements, and bug fixe
 icon: "clock-rotate-left"
 ---
 
+<Update label="v0.34.3" description="September 2, 2026">
+
+**Partner Node Updates**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025): Added the H3 Max model option to MiniMax H3 text-to-video, first-last-frame, and reference nodes
+
+**Performance & Stability**
+* [**TRELLIS remesh memory**](https://github.com/Comfy-Org/ComfyUI/pull/16034): Lowered remesh memory usage in TRELLIS workflows
+
+**Bug Fixes**
+* [**3D preview paths**](https://github.com/Comfy-Org/ComfyUI/pull/16038): Annotate temp paths on Preview3DAdvanced, PreviewGaussianSplat, and PreviewPointCloud so Cloud previews resolve
+
+</Update>
+
 <Update label="v0.34.2" description="August 27, 2026">
 
 **Partner Node Updates**

ja/changelog/index.mdx

@@ -2,9 +2,10 @@
 title: "変更履歴"
 description: "ComfyUI の変更ログ：各リリースの最新機能、改善点、バグ修正を、新バージョンのリリースに合わせて更新。"
 icon: "clock-rotate-left"
-translationSourceHash: 13aa7c8c
+translationSourceHash: 2be39ac3
 translationFrom: changelog/index.mdx
 translationBlockHashes:
+  "v0.34.3": e91d84d1
   "v0.34.2": dd51aaab
   "v0.34.1": f6bfdd8e
   "v0.34.0": ab0efbb1
@@ -124,6 +125,19 @@ translationBlockHashes:
   "v0.3.40": 24608eaf
 ---
 
+<Update label="v0.34.3" description="2026年9月2日">
+
+**パートナーノードの更新**
+* [**MiniMax H3 Max**](https://github.com/Comfy-Org/ComfyUI/pull/16025): MiniMax H3 のテキストから動画、最初と最後のフレーム、リファレンスの各ノードに H3 Max モデルオプションを追加しました
+
+**パフォーマンスと安定性**
+* [**TRELLIS リメッシュのメモリ**](https://github.com/Comfy-Org/ComfyUI/pull/16034): TRELLIS ワークフローでリメッシュ処理のメモリ使用量を削減しました
+
+**バグ修正**
+* [**3D プレビューのパス**](https://github.com/Comfy-Org/ComfyUI/pull/16038): Preview3DAdvanced、PreviewGaussianSplat、PreviewPointCloud の一時パスに注釈を付け、Cloud プレビューが解決されるようにしました
+
+</Update>
+
 <Update label="v0.34.2" description="2026年8月27日">
 
 **パートナーノードの更新**